Choose timezone
Your profile timezone:
FTMI-TBPET-PSMR 2026 Valencia
→
Europe/Zurich
Valencia Hotel Las Arenas
Valencia Hotel Las Arenas
C/ d'Eugènia Viñes, 22, 24, Poblados Marítimos, 46011 Valencia, Spain
Antonio J. Gonzalez Martinez,
Nicola Belcari,
Paul Rene Michel Lecoq
(CERN),
Stefaan Vandenberghe
Description
Molecular imaging is at the forefront of personalised medicine with several technological developments introducing a paradigm shift in the diagnosis, staging and treatment follow-up of patients. Among them, the quest for a higher sensitivity through improved Time-of-Flight PET (TOF-PET) performance and enlarged field of view with Total Body PET (TB-PET) approaches nicely complements a vigorous multimodality effort to combine the merits of PET/SPECT with MRI in single integrated systems.
As the ultimate goal is the same, namely allowing earlier diagnosis of diseases and improving their prognosis, and as several emerging technologies are of common interest for these three research lines, we have recently decided to merge the three dedicated workshop series PSMR (PET/SPECT and MR Multimodal Technologies), FTMI (Fast Timing in Medical Imaging) and TB-PET (Total Body PET) and into a single one to be organised every second year.
The first combined workshop: PSMR-FTMI-TBPET took place in May 2024 in Isola d’Elba, Italy.
We are thrilled to announce the 2d one: FTMI-TBPET-PSMR in May 11-14th 2026 at hotel Las Arenas[1], Valencia (Spain).
We are currently working on finalizing logistic details such as registration fees (and modalities), accommodation options, and the scientific program.
For interested people, this workshop will be immediately followed by the MEDAMI 2026 workshop in May 15-17th 2026 at the Terminal Hub at the Valencia harbour[2], with the theme:
Deployment of Nuclear Medicine Technologies to Low- and Medium-Income Countries
A dedicated indico web page will be open soon for each of these two workshops, with more relevant information. For the time being, book these days in your agenda! We are looking forward to welcoming you in Valencia in May 2026.
For the organizing committee
Antonio J. Gonzalez
Paul Lecoq
Technical program committee
● Nicola Belcari, INFIN Pisa, Italy
● Simon Cherry, UC Davis, Davis, USA
● Maurizio Conti, Siemens, Knoxville, USA
● Antonio J. Gonzalez, I3M (CSIC/UPV), Valencia, Spain
● Andrea Gonzalez-Montoro, I3M (CSIC/UPV), Valencia, Spain
● Paul Lecoq, I3M (CSIC/UPV), Valencia, Spain, and Metacrystal SA., Geneva, Switzerland
● Jae Sung Lee, Seoul National University College of Medicine, Seoul, South Korea
● Robert Miyaoka, University of Washington, USA
● Magdalena Rafecas, Lübeck University, Germany
● Andrew Reader, King's College London, UK
● Kuangyu Shi, Universitätsspital Bern, Switzerland
● Dimitris Visvikis, University of Brest, France
● Stefan Vandenberghe, Ghent University, Belgium
Local Organizing committee
● Antonio J. Gonzalez, I3M (CSIC/UPV), Valencia, Spain
● Paul Lecoq, I3M (CSIC/UPV), Valencia, Spain, Metacrystal SA. and CERN, Geneva, Switzerland
● Marta Freire, I3M (CSIC/UPV), Valencia, Spain
● Fiammetta Pagano, I3M (CSIC/UPV), Valencia, Spain
● Patricia Sanjuan, Turevents & Go, Valencia, Spain
● Daniela Alvarado, Turevents & Go, Valencia, Spain
Contact: Patricia Sanjuan, Daniela Alvarado
-
-
Welcome session
- 1
-
Session 1: FTMI-Overview & ScintillatorsConvener: Paul Rene Michel Lecoq
-
2
Overview of Fast Timing developments for PET
Introduction to FTMI
Speaker: Georgios Konstantinou -
3
Recent developments in the field of scintillators for radiation detectors
Since many decades scintillating crystals have been used for radiation detectors such as high resolution electromagnetic calorimeters and positron emission tomographs. Significant progress has been made in the field of inorganic scintillators in the understanding of their scintillation properties, radiation hardness and production methods over the last 35 years. In addition, many applications also have more and more need for an improved timing resolution. To this purpose many studies have been carried out in the framework of the Crystal Clear Collaboration on the investigation, improvement and exploitation of different processes.
In this contribution we will present selected results of recent research efforts and developments on scintilllators for future detectors.Speaker: Etiennette Auffray Hillemanns (CERN)
-
2
-
Session 2: FTMI-ScintillatorsConvener: Etiennette Auffray Hillemanns (CERN)
-
4
Ultrafast exciton-based scintillation in specifically engineered perovskites
Ultrafast scintillators are critically demanded in a wide range of advanced radiation detection scenarios, including high-speed X-ray imaging, positron emission tomography (PET), time-of-flight (TOF)–based neutron and gamma-ray discrimination, and timing diagnostics at synchrotron radiation facilities. However, the scintillation response of conventional materials is fundamentally constrained by ion-doped luminescence mechanisms, which rely on intrinsically slow electronic transitions.
In contrast, we have proposed and experimentally demonstrated an exciton-based scintillation paradigm, in which radiative recombination originates from electronically excited states with intrinsically ultrafast dynamics. By exploiting diverse excitonic species—including self-trapped excitons in copper halides, hot-exciton emission in organic scintillators, and strongly confined free excitons in two-dimensional perovskites—we establish a new family of scintillators that simultaneously achieve ultrafast response, high light yield, and excellent radiation sensitivity.
In this talk, I will present a comprehensive discussion spanning excited-state mechanism characterization, crystal growth and materials engineering, and application-oriented detector and device design. These results highlight the central role of exciton dynamics in breaking the conventional speed limits of scintillation and open new opportunities for next-generation high-spatiotemporal-resolution radiation detection systems.Speaker: Paul Rene Michel Lecoq -
5
Timing investigation of Tl2LaCl5: a high sensitivity fast scintillator for medical imaging detectors
Radiation detectors play a crucial role across a wide range of applications. Emerging technologies, such as photon-counting computed tomography and advanced single-photon emission computed tomography imaging, demand a combination of high energy resolution, fast scintillation kinetics, and high stopping power. Cerium-doped thallium lanthanum chloride (Tl$_2$LaCl$_5$:Ce, TLC) has recently been introduced as a promising scintillator candidate that combines these properties, making it an attractive material for next-generation radiation detection systems.
We investigate the energy resolution under various excitation energies and the achievable coincidence time resolution (CTR) of two identical crystals coupled to NUV-HD SiPMs. An energy resolution of 4.2% at 511 keV was measured, enabling a clear separation between full-energy deposition and the K-shell escape events. By selecting either 511 keV coincidence events or K-shell escape events, a CTR of 215 $\pm$ 4 ps and 232 $\pm$ 6 ps was obtained, respectively. These values are more than eight times better than expected based on literature values considering scintillation yield and kinetics.
Cramér–Rao lower-bound calculations of the achievable time resolution reveal a strong dependence on the scintillation rise time. The measured timing performance requires a scintillation rise time faster than $\tau_r$ = 2 ns when accounting for prompt Cherenkov photons, or faster than $\tau_r$ = 300 ps in the absence of such prompt photon contributions.
These findings indicate that the scintillation rise time of TLC may be substantially faster than previously reported (4-8 ns), highlighting the importance of early scintillation kinetics and Cherenkov light for timing performance. Combining energy resolution around 4%, competitive timing, and high stopping power, TLC could bridge the performance gap between dense scintillators such as LYSO:Ce and high-resolution materials such as LaBr$_3$.
Speaker: Nicolaus Kratochwil (University of California, Davis) -
6
Integration of a BGO Window Faceplate and high quantum efficiency photocathode in Multi-Anode Microchannel Plate Photomultiplier Tubes
Image quality of positron emission tomography (PET) can be enhanced using time-of-flight (TOF) information between a pair of PET detectors. Coincidence time resolution (CTR) of commercially available TOF-PET scanners is as high as ~200 ps full width at half maximum (FWHM), showing improved PET image quality compared with non-TOF PET images. Theoretically, CTR can improve the signal to noise ratio of PET images as the inverse square root of CTR, indicating that further improved CTR can enhance PET image quality. Moreover, when CTR reaches 30 ps FWHM, the spatial precision calculated from CTR becomes 4.5 mm, which is comparable to the spatial resolution of earlier-generation clinical PET scanners. This implies that PET scanners may no longer need to be cylindrical systems that surround a patient’s body and require image reconstruction processes, because such an ultrafast CTR can directly provide spatial information on the electron–positron annihilation position within the patient, potentially making PET systems more compact and flexible. However, achieving a CTR of 30 ps FWHM remains challenging even with the current state of the art scintillators and photodetectors. Detecting Cherenkov photons instead of scintillation photons has recently attracted attention as a possible approach to achieving the theoretical CTR limitation because Cherenkov photons are emitted on the order of picoseconds and can be several orders of magnitude faster than scintillation photons. On the other hand, the number of Cherenkov photons emitted is significantly less than that of scintillation photons. Therefore, effectively collecting Cherenkov photons while maintaining their excellent timing property is challenging. To address this issue, Cherenkov radiator-integrated microchannel plate photomultiplier tube (CR-IMP, formerly CRI-MCP-PMT) was developed and a CTR of ~30 ps FWHM was experimentally demonstrated. Furthermore, the use of a pair of CR-IMPs enabled the acquisition of two-dimensional cross-sectional images of a phantom filled with positron emitting radionuclide using only TOF information, without a conventional image reconstruction process. However, the demonstration employed two single channel detectors, which are too bulky to construct a practical system using multiple detectors. Hence, the development of position-sensitive multi-anode (MA) MCP-PMTs is the next step towards realizing the reconstruction-free imaging concept. In this study, we present the first fabricated 4×4 multi-anode MCP-PMTs whose window faceplate is replaced with bismuth germanate (BGO), which serves as a hybrid Cherenkov/scintillation material that enables simultaneous measurement of ultrafast CTR and energy information, which require clinical systems. This study focuses not only on the integration of BGO into an MA-MCP-PMT structure but also on the implementation of a super bialkali photocathode for enhancing probability of Cherenkov photon detection with high quantum efficiency (QE). At this stage, the timing performance of the detector remains unoptimized. The developed detector has an active area of 23 × 23 mm2 with the channel readout pitch of 5.75 mm. A maximum QE is 27.6% at 340 nm is achieved owing to the implementation of the super bialkali photocathode. In addition to the QE curve, in this presentation, we will show preliminary evaluations of the MA-BGO-IMP, such as gain, dark count rate, single photon response, etc.
Speaker: Dr Ryosuke Ota (UC Davis) -
7
Cherenkov Timing in BGO and Rise Time Classification with FastIC+ ASIC
The ultrafast emission of Cherenkov photons offers a route to enhance the coincidence time resolution (CTR) of time-of-flight positron emission tomography (TOF-PET) detectors. Although the Cherenkov yield under 511 keV excitation is low, these photons increase the signal onset steepness and thus the achievable timing precision. Among common scintillators, bismuth germanate (BGO) emits more Cherenkov photons than lutetium-based crystals, but its slow scintillation component leads to large event-by-event variations in signal shape. Measuring the rise time at the signal onset, obtained from the time difference between two thresholds, allows classification of events by steepness and time-offset correction to improve CTR. While this principle has been demonstrated with dedicated fast electronics, scalable implementation remains a challenge. The new FastIC+ ASIC provides an integrated rise-time measurement correlated with CTR in BGO scintillators. We show event classification using rise time to mitigate time offsets and to improve the CTR by 32%, from 577 to 395 ps FWHM with 3×3×15 mm³ polished BGO with ESR on 3×3 mm² FBK NUV-MT SiPMs biased at 49 V. This work marks the first steps toward a scalable approach to exploiting BGO's mixed Cherenkov–scintillation emission in TOF-PET imaging.
Speaker: Francis Loignon-Houle (Instituto de Instrumentación para Imagen Molecular (I3M), CSIC—UPV)
-
4
-
Lunch Break
-
Session 3: FTMI-Photodetectors: Session 3Convener: Alberto Gola
-
8
Sponsor Keynote: Broadcom’s Silicon Photomultiplier: Technology, Performance, Status and Roadmap
This presentation will provide an overview about Broadcom’s SiPM and photodetection technologies including NIR and NUV-SiPM. Focus will be put on the initial results of the newly developed NUV-DJ SiPM technology, showing a significant increase of the PDE above 420nm compared to the NUV-MT technology, while excellent NUV-sensitivity from 420nm down to 250nm is maintained. At 500nm the PDE was improved from 52% to 65% while keeping the a high dynamic range by a SPAD pitch of 40µm and an overall reduced contribution of correlated noise.
The high PDE combined with low noise are key to best-in-class performance in CTR for LYSO-based TOF-PET and extends the application range to other green emitting scintillators like GaGG and enables BGO based Chrerenkov TOF-PET. Results of the initial characterization of NUV-DJ will be summarized and presented.
Additionally, an overall outlook including the development of a 3x3mm2 HDR-NUV-MT SiPM with SPAD pitch smaller than 20µm and a TEC-package targeting applications requiring high dynamic range or extremely low dark count rates will be presented.Broadcom’s NUV SiPMs find application in TOF-PET, Cherenkov-PET, radiation spectroscopy, photon-counting X-ray detection & PCCT, flow cytometry, and time-gated fluorescence.
Keywords: TOF-PET, Cherenkov, radiation spectroscopy, PCCT, Fluorescence DetectionSpeaker: John Murphy -
9
Sponsor Keynote: Advancements in Photon Counting Technologies: Silicon SPAD & MPPC (SiPM) Developments at Hamamatsu Photonics
Silver sponsor talk
Speaker: Luigi Ghezzi -
10
TSV-enabled segmentation of FBK fast-timing SiPMs for 2.5D and 3D integrated ToF-PET modules
Next-generation medical imaging modalities, particularly high-resolution time-of-flight positron emission tomography (TOF-PET), require photodetectors that combine excellent timing precision, high spatial granularity, and compatibility with advanced system integration. To address these needs, Fondazione Bruno Kessler (FBK) is developing silicon photomultipliers (SiPMs) specifically conceived for 2.5D and 3D integration with dedicated readout electronics through customized through-silicon via (TSV) solutions.
By overcoming the limitations of traditional wire bonding, this TSV-based approach enables high-density interconnections and finer detector segmentation, both of which are essential for precise clinical imaging. At the same time, it improves signal integrity by reducing parasitic output capacitance per channel, thereby preserving fast signal transitions and supporting the excellent coincidence time resolution (CTR) required in modern PET systems. The feasibility of this segmentation strategy is confirmed by the timing performance of micro-SiPM structures: experimental characterization shows that the single-photon time resolution (SPTR) remains substantially unchanged from a 7 × 7 SPAD matrix up to a 1 × 1 mm² active-area SiPM, with values close to those of a single 50 μm SPAD (≈25 ps FWHM) under blue-light excitation and high-frequency readout.
This integration strategy is currently implemented on FBK’s NUV-HD-MT (Near-Ultraviolet High-Density Metal-in-Trench) SiPM technology, which is optimized for timing applications and features an excellent SPTR, a photon detection efficiency (PDE) of about 65%, a dark count rate below 80 kcps/mm², and a direct crosstalk probability below 3% at 10 V excess bias. The same approach is expected to be extended to the more advanced NUV-DJ technology, recently demonstrated by FBK with even higher PDE while preserving excellent timing performance.
Two TSV solutions are under development. The first is a glass-less TSV, which supports segmentation below 0.5 mm pitch and removes the glass carrier wafer after thinning, enabling direct coupling between the scintillator and the SiPM and avoiding intermediate materials that could degrade timing performance or radiation tolerance. The second is the Single-Cell Access (SCA) TSV, which enables interconnection down to the microcell level. In this approach, the SiPM microcells are electrically isolated and can be regrouped on the backside through a redistribution layer into micro-SiPMs, such as 3 × 3 or 6 × 6 cell clusters, for applications requiring extremely fine granularity.
Comprehensive integration tests were carried out to validate the TSV process and backside interconnection. Laser-assisted ball deposition demonstrated nearly 100% yield at 500 μm pitch, with a resistance below 3 Ω per ball. A representative system-level application is the PETVision project, where glass-less TSV-integrated 12 × 6 mm² SiPM dies, each embedding a monolithic 2 × 4 array of 3 × 3 mm² channels, will be combined with a dedicated ICCUB ASIC in a 2.5D architecture to build 50 × 50 mm² photon detection modules with high fill factor and seamless tiling for next-generation TOF-PET.
Speaker: Carina Trippl -
11
Developments in fabrication and design flexibility of amorphous silicon microchannel plates
The technology of amorphous silicon-based micro-channel plates (AMCPs) uses a versatile approach to stack hydrogenated amorphous silicon in thicknesses up to 100 μm, using plasma-enhanced chemical vapor deposition (PECVD) and etch microchannels of diameter below 2 μm every 5 μm, on a hexagonal pattern, by deep reactive ion etching. With gains up to 1000 for reverse bias less than 300 V, the design is possible to build on arrays of sensitive transimpedance amplifiers, integrated in read-out circuits with high channel density, such as for instance 15 by 15 μm. This ensures single photon detection with quantum efficiency up to 30% and noise rates below 50 Hz/cm², if combined with state of the art photocathodes such as the hi-QE from Photonis. Current fabrication runs focus on direct integration of the stack on ASICs, to be integrated with the photocathode over the next months. Measurements show timing response with jitter below 10 ps, making this high-precision photodetector ideal for radiation detectors combining ultra-fast photon emission and low light yield, such as Cherenkov.
Speaker: Georgios Konstantinou (EPFL) -
12
Experimental Evaluation of a Low-Noise Ultra-Fast SiPM Time-of Flight Analog Readout Chip (LUSTAR)
The precision to which the arrival time of 511 keV photons are able to be known directly affects the signal-to-noise-ratio (SNR) for reconstructed images in time-of-flight positron emission tomography (TOF-PET). Currently, significant research is ongoing to push the full-width-half-maximum (FWHM) coincidence time resolution (CTR) for TOF-PET systems well below 100 ps towards the few-10’s-of-ps regime. In this work, we designed an application specific integrated circuit (ASIC) in a commercial 65nm CMOS process to translate ultra-fast timing performance currently demonstrated in state-of-the-art bench-top measurements into a fully integrated scalable topology for large scale time-of-flight positron emission tomography (TOF- PET) systems. We targeted a scalable, low-noise, ultra-fast silicon photomultiplier (SiPM) analog front-end that maintains excellent SNR for single photon response while achieving ≤35 ps FWHM electronic noise jitter for single photon detection with 4x4 mm2 Broadcom AFBR-S4N44P014M Near-ultraviolet metal-trench SiPMs. This “analog” chip uses low-voltage differential signaling (LVDS) transmitters to drive discriminator data to an external time-to-digital converter (TDC). We present experimental evaluation of achievable single photon time resolution and CTR with the first prototype test chip.
Speaker: Joshua Cates (Lawrence Berkeley National Laboratory)
-
8
-
Coffee Break
-
Session 4: FTMI-ElectronicsConvener: Joshua Cates
-
13
Approaching Sub-100 ps CTR in TOF-PET Scanners with FastIC+
In recent years, several research groups have demonstrated the potential of time-of-flight positron emission tomography (TOF-PET) by achieving coincidence time resolutions (CTR) below 100 ps in the laboratory. These results, however, rely on small scintillator crystals -favoring timing performance at the expense of detection efficiency- and/or bulky, high-power readout electronics. A realistic system-level demonstration requires, at least, $\sim$20 mm thick scintillator crystals and dedicated readout electronics capable of preserving the intrinsic timing performance of the detectors while maintaining low power consumption and high channel density.
We have developed FastIC+, a custom application-specific integrated circuit (ASIC) designed for fast-timing applications and optimized for the readout of high-gain photodetectors such as silicon photomultipliers (SiPMs). FastIC+ integrates a low-noise analog front-end with an on-chip Time-to-Digital Converter (TDC) with 25 ps binning. The power consumption is ~12 mW per channel.
The performance of FastIC+ was evaluated in a TOF-PET laboratory experiment using SiPMs coupled to scintillation crystals of different sizes and materials. With small LGSO crystals we achieved, for the first time, a CTR below 100 ps using full-ASIC readout, including on-chip digitization. To approach system-level conditions, we further evaluated the ASIC with dual-ended readout TOF-PET detectors equipped with 20 mm thick LYSO crystals. Despite a non-optimized setup, a CTR of 120 ps was achieved, opening a clear path toward sub-100 ps performance in realistic scanner configurations. In addition, to enable scalable system integration, we are developing dedicated 64- and 256-channel electronic modules for TOF-PET scanners.
Speaker: David Mazzanti Tarancon (Institute of Cosmos Science, University of Barcelona) -
14
Initial development of a modular FPGA-based DAQ for the PeTVISION ultrafast PET system
The development of the data acquisition system (DAQ) is one of the key technological contributions of I3M within the PeTVISION project, a collaborative international effort involving FBK, ICCUB, IJS, ONCOVISION, Yale University, and MRI-TUM Klinikum rechts der Isar. Within this framework, I3M, with support from ONCOVISION, is responsible for the FPGA-based nuclear DAQ, including the design of scalable concentrator boards (PETV_CB) and the DAQ architecture required to support a highly modular PET system, while also contributing to the development of the system’s synchronization and coincidence detection board (PETV_TRG), led by ONCOVISION. The overall architecture is intended to scale from validated elementary building blocks up to the complete system.
The proposed DAQ is based on Xilinx Artix-7 FPGAs and is designed to support a PET system with more than 18,000 readout channels. In the initial implementation, the front-end readout is based on the FastIC+ ASIC, an 8-channel chip with integrated TDC that provides digitized timing information and transmits nuclear data through a simplified Aurora 64B/66B single-lane link operating from 80 Mb/s up to 1.28 Gb/s. The DAQ architecture foresees one PETV_CB concentrator board handling 4 photodetector modules, each of them connected through 16 Aurora links for ASIC data readout. In addition, the board provides 4 I2C buses for ASIC control and up to 16 extra I2C devices for sensing and automated supervision of the readout system.
The work presented here focuses on the preliminary design and validation of the fundamental DAQ building blocks. Although full validation will only be achieved once the complete scalable architecture is assembled, the results obtained so far already demonstrate the feasibility of the proposed approach. First, we validated the read and write access to FastIC+ internal registers through I2C, confirming correct low-level control of the ASIC. Second, we validated the reception of nuclear data transmitted from a FastIC+ ASIC to an Artix-7 FPGA prototype through Aurora 64B/66B, which represents a major milestone toward multichannel acquisition. The analysis of the received FastIC+ stream also confirmed the relevance of event packets, delimiter blocks, idle blocks, and monitoring-related packet structures required for robust receiver development.
A further important preliminary result is the demonstration of high-throughput data transfer between the FPGA prototype and a host computer, reaching sustained readout rates on the order of 1 GB/s. Taken together, these results show that the combination of ASIC control through I2C, Aurora-based nuclear data reception, and high-speed data extraction can be coherently integrated into a single FPGA-based DAQ platform. While still preliminary, this work establishes the validated basic blocks on which the future multiboard, multichannel, and full-scale PeTVISION DAQ system will be built.Speaker: Dr Noriel Pavon Hernandez (Institute for Molecular Imaging Instrumentation (I3M)) -
15
Channel-Level Validation of a Calibrated FPGA TDC for Fast-Timing PET
We present a fast-timing-oriented digital readout architecture for TOF-PET detector modules based on the HRFlexToT front-end ASIC and a calibrated multi-channel FPGA Time-to-Digital Converter (TDC). Following the earlier UTOFPET detector developments, the firmware and timing back-end have been redesigned from scratch with specific emphasis on timing digitization robustness, channel-level validation, and scalability. The proposed TDC is based on a tapped-delay-line architecture with Nutt interpolation and offline calibration through histogram-derived bin-width estimation and characteristic-curve reconstruction. Measurements on Cyclone 10 devices show that placement-aware implementation is essential to control non-uniform bins induced by clock-region boundaries, reducing ultra-bins from about 200–300 ps to 30–120 ps. In channel-level tests on Cyclone 10, a 256-tap, 64-bin configuration with 2 ns clock period yields an LSB of 35.1 ps, an ultra-bin of 110.5 ps, and 40.2 ps rms timing precision, while resource estimates indicate compatibility with 64-channel integration on the target device. In parallel, the timestamp aggregation and USB3 readout chain have been validated up to 1 Gbps, corresponding to 32 Msps total throughput or about 500 ksps per channel over 64 channels. These results establish a validated digital timing platform for scalable fast-timing PET instrumentation.
Speaker: Giancarlo Sportelli -
16
TEMPOROC: A 64-channel SiPM read-out chip for time-of-flight measurement
TEMPOROC3 is a 64-channel SiPM read-out chip designed in TSMC130 technology. The ASIC aims at measuring the energy with 10-bit ADC, time-of-arrival (TOA) and time-over-threshold (TOT) with 40 ps bin TDC. The analog front-end offers a dynamic range up to 500 pC. The dual discriminator allows either to measure two TOA at different thresholds and at either rising or falling edge. Measurements in coincidence between 2 ASICs have shown TDC time resolution as low as 25 ps RMS (CTR). The dual-gain charge measurement integrates the input signal over a tunable range from 20 to 300 ns. The 64 channels are divided in 4 clusters, each outputting data at 320 Mbps. The clusters can be easily synchronized to stream all channel data whenever a channel from the matrix has triggered. The number of triggered channels (per cluster) required to launch the data serialization is programmable as well as the window duration for accepting new triggers. At 450 kEvents/s (with 1 TDC + 1 ADC per channel), the power consumption is about 8 mW per channel. TEMPOROC3 is a very versatile read-out ASIC with numerous configuration parameters that can find applications in for example Compton camera, LIDAR or also ToF-PET.
Speaker: Julien Fleury (Weeroc)
-
13
-
Welcome reception
-
-
-
Session 5: FTMI-Detectors/MethodsConvener: Andrea González Montoro
-
17
Improving TOF and DOI capabilities of pixelated PET modules using High-Frequency Readout
Low-dose imaging and small-animal positron-emission tomography (PET) require detectors with both high sensitivity and high spatial resolution. In this study, we investigate the performance of high-density scintillators, such as Bismuth Germanate (BGO) and LYSO:Ce, with multi-channel, low-noise, low-power, high-frequency (HF) readout electronics to enhance the Time-of-Flight (TOF) and Depth-of-Interaction (DOI) performance of pixelated detectors. High-frequency SiPM readout has demonstrated significant improvements in coincidence time resolution (CTR) by enabling the detection of ultrafast optical-photon production, such as Cherenkov emission in BGO and fast scintillation in LYSO:Ce.
Two DOI-encoding detector designs are compared in this work: a double-sided readout and a single-sided readout employing a light guide to achieve DOI encoding through light sharing of laterally depolished crystal matrices. The detector modules are read out using a custom sixteen-channel low-noise HF electronics board capable of setting a low leading-edge threshold for the detection of the earliest photons produced. In combination with metal-in-trench Broadcom SiPMs, which provide high photon detection efficiency in the relevant wavelength range, this approach enables improved timing and DOI resolution.
Experimental results demonstrate that matrices of 20 mm long LYSO:Ce crystals can achieve a DOI resolution of approximately 2 mm FWHM and a CTR of 130 ps FWHM using the light-sharing method. BGO crystals show a DOI resolution above 7 mm FWHM with the same method; however, this improves to 4.5 mm FWHM when a double-sided readout is employed, which also provides a CTR of 254 ps FWHM. The double-sided readout approach will be further investigated using LYSO matrices, where improved performance in terms of both timing and DOI resolution is expected.Speaker: Dr Giulia Terragni (CERN) -
18
Maximum Likelihood Timing Estimation in Monolithic Scintillators using the PETsys TOFPET2 ASIC
Introduction: Monolithic scintillators represent a cost-effective alternative to pixelated detector designs in the development of time-of-flight (TOF) PET scanners. However, achieving optimal coincidence time resolution (CTR) in monolithic blocks is limited by the variance in photon arrival times and the complex optical transport across the detector volume. This work evaluates a timing reconstruction method based on maximum likelihood estimation (MLE) using the PETsys TOFPET2 ASIC, which provides individual timestamps for each SiPM channel in the array.
Methods: We evaluated a 32x32x10 mm³ LYSO monolithic crystal wrapped with white reflector coupled using optical grease to an 8x8 Broadcom NUV-MT SiPM array with a 4 mm pitch. The signals were processed using the PETsys TOFPET2 Evaluation Kit. To implement the MLE algorithm, reference probability density functions (PDFs) of the optical transport were derived using a coincidence setup with a Na-22 point source and a 3x3x4 mm³ LYSO pixel as a timing reference to perform a gamma beam scan. The MLE estimator was compared against conventional timing methods, including skew-only and combined skew and time-walk corrections. Additionally, the performance was compared against the numerical average of a varying number of SiPM timestamps and energy-weighted average methods.
Results: The CTR performance was characterized for all evaluated methods. The CTR FWHM obtained using the skew-only correction was 505 ps. The numerical average of N-SiPMs yielded 466 ps (N=2), 534 ps (N=3), 716 ps (N=4), and 1079 ps (N=5), while the energy-weighted average reached 452 ps (N=2), 470 ps (N=3), 548 (N=4), and 688 ps (N=5). The MLE-based timing method achieved a CTR FWHM of 423 ps.
Conclusion: The implementation of an MLE-based timing estimator with the PETsys TOFPET2 ASIC improves the coincidence time resolution of monolithic LYSO detectors. By modeling the optical response through experimental PDFs, the MLE method compensates for the timing degradation inherent in non-segmented crystals. These results demonstrate that the application of statistical estimators to individual SiPM timestamps can align the timing performance of monolithic scintillators with that of more complex pixelated geometries, supporting the use of monolithic designs in high-sensitivity TOF-PET systems.Speaker: Ms Helena Segarra (Consejo Superior de Investigaciones Científicas (CSIC)) -
19
Improving Coincidence Time Resolution using Maximum Likelihood Estimation in Dual-Side Readout Scintillators
Introduction: High-resolution depth-of-interaction (DOI) and coincidence time resolution (CTR) are critical factors for improving the performance of next-generation Time-of-Flight (TOF) PET scanners. While dual-side readout (DSR) configurations provide DOI information, the precise estimation of the interaction time remains a challenge due to optical transport delays and the stochastic nature of photon detection. This work presents a timing method based on maximum likelihood estimation (MLE) specifically designed for DSR detectors to optimize coincidence timing resolution.
Methods: The study includes Monte Carlo simulations, including scintillation and Cherenkov light production, and experimental measurements using 4x4x20 mm3 LYSO and BGO crystals coupled to Broadcom NUV-MT SiPMs on both ends. The MLE algorithm utilizes reference probability density functions (PDFs) that characterize the optical transport and SiPM response at different DOI layers. For the experimental data, these PDFs were acquired using a Na-22 point source and coincidence measurements with a 4x4x5 mm3 LYSO crystal. The timing reconstruction process first performs a DOI estimation for each crystal and subsequently estimates the interaction time.The CTR performance was evaluated using two identical DSR detectors by processing four timestamps per event using MLE algorithm and compared against single-side readout (front-only and back-only) and the numerical average of both SiPMs of each crystal.
Results: The CTR FWHM (FWTM) values, in ps, for LYSO simulations were 281.7 (517.0) for back SiPM, 265.2 (524.5) for front SiPM, 155.7 (291.0) for numerical average, and 152.7 (291.6) for MLE. For BGO simulations, values were 267.0 (811.6) for back SiPM, 228.6 (697.9) for front SiPM, 124.5 (621.7) for numerical average, and 119.4 (286.8) for MLE. Experimental LYSO measurements yielded 159.3 (541.2) for back SiPM, 143.8 (678.8) for front SiPM, 137.7 (435.5) for numerical average, and 143.9 (520.9) for MLE. Experimental BGO values were 1332.5 (3457.9) for back SiPM, 1325.5 (3211.5) for front SiPM, 1220.4 (2963.8) for numerical average, and 944.2 (2050.5) for MLE.
Conclusion: The integration of MLE with DSR detectors improves timing estimation by incorporating the statistical information of both scintillation and Cherenkov photons. This approach compensates for the degradation caused by DOI effects and optical transport delays. The results demonstrate that while the difference of averages improves the FWHM, the MLE method is necessary to effectively constrain the temporal distribution. This is particularly evident in BGO crystals, where the MLE utilizes the fast Cherenkov photon contribution to narrow the timing peak and suppress the broad temporal tails characteristic of BGO scintillation. These findings indicate that MLE is an effective tool for achieving high-resolution coincidence timing in TOF-PET systems utilizing long scintillation crystals. Current experimental results are limited by the low sampling rate of the acquisition system. Further refinements using higher-speed electronics are currently underway to minimize this instrumental jitter.Speaker: CLAUDIA FERNANDEZ BORREGUERO -
20
Benchtop Characterization of a 64-Channel Scintillation Photon Counting TOF-PET Monolithic Detector
We characterized the performance of a time-of-flight positron emission tomography (TOF-PET) detector based on our scintillation photon counting technique. The detector consists of a 32.2 × 32.2 × 20 mm3 monolithic Fast LGSO crystal, an 8 × 8 array of Broadcom NUV-MT silicon photomultipliers (SiPMs), and a 64-channel low-noise, high-frequency (LNHF) electronics architecture. This detector design enables unique timestamping of multiple early arriving scintillation photons, which greatly improves the precision of interaction time estimation. Continuous 3D event positioning capability is another benefit of monolithic detectors, allowing correction for the transit time of scintillation photons. Timing and energy signal waveforms were sampled using two CAEN V1742 and one V1740 digitizers. For 3D event positioning, we performed 1D calibration and used convolutional neural networks (CNNs) to estimate X, Y, and Z from 8 × 8 energy signal amplitudes. The measured positioning resolution was 1.93 × 1.92 × 2.32 mm3. For event timing, maximum-likelihood interaction time estimation (MLITE) was applied to the 8 × 8 timing signal leading edges, achieving a preliminary coincidence time resolution (CTR) of 141 ps with a reference detector. The latest results using an optimized readout configuration will be presented.
Speaker: Seungeun Lee -
21
Cherenkov Photon Counting, Labeling, and Multi-Photon Timing Estimation for sub-100 ps CTR BGO TOF-PET Detectors
We have developed a bismuth germanate (BGO) Cherenkov and Scintillation photon counting time-of-flight positron emission tomography (TOF-PET) detector which can count and provide unique timestamps for detected optical photons. This allows 511 keV photon time of interaction estimators to be derived from the prompt Cherenkov light without influence from scintillation light, which has a much higher variance from time of detection. This detector concept is also capable of three-dimensional positioning of 511 keV photon interactions, allowing for position and depth-of-interaction dependent optical transit time skew corrections to be applied to the prompt estimators. Information from this detector concept can be used to classify each detected optical photon as Cherenkov or scintillation. In this work, we show how labeling each detected photon as Cherenkov/scintillation and only incorporating prompt information into a timing estimator that averages timestamps for events with more than one detected Cherenkov photon significantly improves both achievable coincidence time resolution with BGO and also the TOF kernel shape by significantly reducing the long tails from time pickoff on scintillation light. We constructed a 16x16x15 mm3 BGO photon counting detector prototype and evaluated its performance in coincidence with a fast reference detector. Incorporating the information from multiple labeled Cherenkov photons improved CTR from 159 ps to 137 ps FWHM, and the long scintillation tail was reduced to a small fraction of the total coincidence distribution. When considering all events with at least one Cherenkov photon detected, CTR was 122 ps FWHM with a Gaussian kernel shape. CTR improved to ≤100 ps for events containing 2 or more detected Cherenkov photons. We present methods for these methods, along with a new detector concept with high Cherenkov photon detection efficiency to enable ≤100 ps CTR in a BGO TOF-PET detectors that also have high 511 keV photon detection efficiency.
Speaker: Joshua Cates (Lawrence Berkeley National Laboratory)
-
17
-
Coffee Break
-
Session 6: FTMI-SystemsConvener: Fiammetta Pagano
-
22
A metascintillator-based, ultra-ToF enabled, brain PET scanner simulation
Metascintillator systems composed of LYSO:Ce and MAPbBr3 perovskites were recently measured in the laboratory with high end electronics. The timing behavior of this metascintillator rendered two dominant subsets: 44% of fast, perovskite-led and shared events at 46 ps and 56% of slow, LYSO:Ce events at 110 ps. With such coincidence timing resolution (CTR), ultrafast ToF regimes are reached, and this offers an opportunity to better understand the effect of such CTR on final image quality. The possibility of adapting this configuration to semi-monolithic metascintillator also offers very good depth-of-interaction discrimination, at 3 mm. To benchmark this detector architecture, a brain PET scanner architecture was developed in GATE, with a Hoffman brain phantom as the leading approach. A series of reconstruction approaches, for instance using the fast subset as priors or weighing in on the different subsets, are to take place. Data will become available as an open invitation to the reconstruction community to experiment with, to demonstrate the potential of new generation high end detectors for medical imaging.
Speaker: Georgios Konstantinou (EPFL) -
23
Detector-Wise Learning for Scalable In-System PET Timing Calibration
Accurate detector timing calibration is essential for time-of-flight positron emission tomography (TOF-PET), as coincidence time resolution (CTR) determines localization accuracy along the line of response (LOR) and affects image signal-to-noise ratio (SNR). Conventional calibration methods such as time-skew and time-walk correction rely on low-dimensional parametric models applied at the detector or channel level, which are often too restrictive to capture complex timing distortions arising from scintillation light transport, SiPM response, and readout electronics. Recent machine learning approaches address this limitation by regressing time differences directly from detector pair signals; however, they do not explicitly decompose timing errors into detector-wise contributions, limiting interpretability and preventing direct detector-level calibration.
We propose a structured learning framework that formulates timing calibration as detector-wise interaction-time offset prediction under coincidence-level supervision. A shared neural network estimates offsets for individual detector interactions from normalized timestamp matrices, photon-count distributions, local interaction position, deposited energy, and a learned detector embedding. Coincidence time differences are reconstructed as the difference of predicted offsets with reintroduced minimum timestamp terms, enabling parameter scaling linear in the number of detectors while sharing statistical strength across sensing elements. Training targets are derived geometrically from known annihilation positions with light-propagation corrections in LYSO.
The method was evaluated on a semi-monolithic prototype PET system comprising 24 detector modules arranged in three axial rings. Each module consists of two LYSO slab arrays read out by digital photon counters, providing $12\times12$ photon-count matrices and $6\times6$ timing channels. A total of 221 million coincidences were acquired from 450 $^{22}$Na point-source positions. Five-fold cross-validation with source-position-level splits (60/20/20) ensured evaluation on spatially unseen positions.
Across three energy windows (470--550 keV, 430--590 keV, 350--650 keV), an embedding-conditioned residual multilayer perceptron (1.83 M parameters) consistently outperformed classical calibration. For the 470--550 keV window, CTR improved from $475.16 \pm 0.57$ ps (time-skew) and $459.33 \pm 0.63$ ps (time-walk) to $422.52 \pm 0.54$ ps. Similar improvements were observed across wider energy windows, with expected degradation from inclusion of lower-energy events exhibiting higher intrinsic timing variance.
A lightweight residual MLP variant (0.89 M parameters) achieved comparable performance with only minor CTR degradation, enabling deployment on resource-constrained hardware. Feature ablation shows that, beyond timestamps, light-spread information provides the strongest auxiliary signal, positional features yield additional improvements, and deposited energy contributes only marginally. Excluding 10\% of detector pairs during training did not degrade performance on the held-out pairs at test time. This indicates that the proposed learning system effectively decomposes errors into consistent single-interaction offsets.
Our results show that the proposed learning system outperforms classical time-skew and time-walk calibration while decomposing coincidence timing errors into detector-wise interaction offsets under coincidence-level supervision, enabling statistically efficient learning, linear parameter scaling with detector count, and straightforward integration of additional signal features.
Speaker: Julian Thull (Chair of Imaging & Computer Vision, RWTH Aachen University) -
24
Comparison of digital and analog PET systems: performance and image quality
Background and aims:
Digital PET systems based on silicon photomultipliers (SiPM) provide improved spatial and timing performance compared with analog photomultiplier tube (PMT) detectors. This study compares a digital GE Discovery MI (20 cm axial FOV, 4 rings) with an analog GE Discovery IQ (26 cm axial FOV, 5 rings) through standardized quality control tests [1].Methods and materials:
Both systems were evaluated using the standardized quality control tests PET02 (spatial resolution), PET03 (sensitivity), PET04 (NECR peak), PET06 (image quality), and PET08 (timing). Analogous phantoms and acquisition parameters were used.Results:
Spatial resolution was superior in the digital PET due to its smaller SiPM detector elements, with both systems converging only at large radial distances. Sensitivity was higher in the analog PET, mainly due to its larger axial FOV and higher number of detector rings. At higher activity concentrations, both systems showed similar true count rates near 5 kBq/mL. The digital PET exhibited a substantially higher NECR peak at higher activity levels and maintained performance across a wider concentration range, despite having slightly higher scatter fractions. Image quality demonstrated greater contrast recovery in the digital PET, although local uniformity around high contrast regions was slightly reduced, likely due to its shorter axial FOV requiring more bed positions. The smaller number of detector rings also increases total acquisition time. However, timing resolution strongly favored the digital system, enabling effective TOF correction.Key words:
Digital PET; Analog PET; NECR; Sensitivity; SiPM; Timing; Image quality.[1] SEFM, Protocolo de Control de Calidad de la Instrumentación en Medicina Nuclear. 2020.
Speaker: Ms Andrea González Rodríguez (Hospital Universitario de Cuenca) -
25
High Sensitivity Multimodal DOI-TOF PET and SPECT for Proton Therapy and Diagnostic Imaging
Overview: We report on ongoing research and development of configurable multi-modal PET and SPECT imaging and dosimetry technology that shows great promise for range verification in proton therapy, imaging of injected radio-pharmaceuticals, or as high resolution insert cameras enhancing MRI. With a large solid angle geometry configuration, boosted by the depth of interaction (DoI) and time of flight (ToF), the sensitivity of such instruments reaches unprecedented high levels thus enabling high-resolution low-dose imaging. It can characterize the FLASH irradiations micro-environment necessary to elucidate the effect or be used for mapping dementia and Alzheimer’s disorders. The versatility of our family of instrument designs presents an unparalleled complete nuclear imaging diagnostics system.
Proton therapy: Improving radiation therapy could reduce common post-treatment complications and significantly boost the overall treatment effectiveness. Radiotherapy mostly relies on gamma rays that affect both cancerous and surrounding healthy tissue, leading to side effects and toxicity. Proton beam therapy, in contrast, offers precise energy deposition, allowing for targeted treatment with little collateral damage. Moreover, therapeutic proton beams, unlike gamma beams, emit secondary radiation that can be monitored in real time if suitable instruments are employed. However, while a lot of effort is involved in treatment planning, the in vivo proton range verification– confirming that the beam is delivered as intended – remains underdeveloped. Our patented ideas directly address this unmet need of proton therapy by advancing in-beam cameras for the in vivo diagnostics that can lead to real-time irradiation verification and adaptive treatment protocols that would considerably improve treatment precision and therapy outcomes.
Importantly, the recently discovered but not-yet-understood FLASH effect – better relative sparing of healthy tissues compared to cancer tissues when treated by very high-dose and ultra-short beams, has the potential to revolutionize radiation oncology. However, any future clinical use of FLASH therapy will require much more accurate delivery with the in vivo feedback diagnostics and new therapies with just a few beam fractions are expected to lead to a much higher patient throughput. To implement these beam therapy monitoring ideas, we propose small, configurable, low-cost PET and SPECT scanners. Unlike large footprint commercial units, our instruments can lead to a much broader use of nuclear imaging and dosimetry in clinics and in research labs.
Versatility: We enhance capabilities of an instrument realized as ToF PET for Proton Therapy by a US-Portugal consortium. A new double-ended readout, necessary for DoI, and gamma collimation extend the technology to enable prompt gamma imaging (PGI) using single photon emission computed tomography (SPECT) during the beam spill. Our instruments respond to wide-ranging needs not only in proton therapy or oncological imaging and dosimetry but also in neurological clinical diagnostics and in research. The ultimate objective is to develop cost-effective imaging modules suitable for cost-effective clinical integration and research applications, delivering an unparalleled tool for both proton therapy, and radiation biology research. We will present how these concepts have been so far validated in both conventional and FLASH proton beams, and are now being established at a commercial level.Speaker: Karol Lang (University of Texas at Austin)
-
22
-
Lunch Break
-
Session 7: FTMI-SytemsConvener: Dennis Schaart
-
26
Design and Prototyping of a Sub-250 ps Cardiac TOF-PET System
Positron emission tomography (PET) is the internationally recognized gold standard for the accurate quantification of myocardial flow reserve (MFR) and metabolic status, conventional oncology-centric systems struggle with the high-frequency motion and small-target nature of cardiac imaging.
To facilitate dynamic quantification and functional imaging, we are developing a high-performance, cardiac-dedicated TOF-PET system. The target technical specifications include $\text{sub-}250 \text{ps}$ TOF resolution, $\text{sub-}1.1 \text{mm}$ spatial resolution, and $\text{sub-}3 \text{mm}$ depth-of-interaction (DOI) resolution. The system further integrates both Electrocardiogram (ECG) gating and exercise stress equipment, aiming to achieve multi-dimensional cardiac functional imaging, such as simultaneous perfusion-metabolism assessment, under exercise stress conditions.
These preliminary results demonstrate the system's potential to provide high-resolution, motion-corrected cardiac functional imaging for clinical diagnostics.Speaker: Qiyu Peng -
27
Initial Human Static and Dynamic FDG Imaging with the SmartBrain Wearable PET
Introduction
Wearable brain PET could extend molecular brain imaging beyond conventional fixed scanners by enabling time resolved acquisition in a more flexible form factor. A first technical requirement is to demonstrate interpretable human static images, preservation of relative regional uptake patterns versus a clinical scanner, and stable extraction of dynamic ROI level information.
Methods
Initial human static and dynamic FDG studies were performed with the SmartBrain wearable PET system. For static evaluation, one volunteer received a single 18F-FDG injection of 6.1 mCi. GE DMI PET was acquired at 53 min post injection with an 8 min acquisition, and SmartBrain PET was acquired at 147 min post injection with a 60 min acquisition. Both datasets were registered to MNI space. ROI analysis used the Neuromorphometrics atlas with within-scan whole-brain normalization after exclusion of CSF, ventricular, and vascular labels. For dynamic evaluation, a separate healthy volunteer underwent a 60 min SmartBrain scan. A 2.6 mCi 18F-FDG dose was injected during the first 30 s after scan start. Dynamic framing was 10 s × 12, 30 s × 6, 60 s × 5, and 300 s × 10, yielding 33 frames over 60 min. Representative ROIs included calcarine cortex, middle frontal gyrus, precuneus, putamen, and cerebral white matter.
Results
Static results showed that although the wearable system had shorter axial coverage, intracranial metabolic distribution remained visually interpretable. ROI pattern analysis across 118 atlas-defined regions with a minimum voxel threshold of 200 demonstrated strong agreement between SmartBrain and GE DMI in relative regional uptake pattern, with Spearman = 0.816 and Pearson = 0.843. Bilateral mean ROI values in representative deep gray matter regions, including thalamus, caudate, putamen, and pallidum, also showed good agreement between the two systems. Dynamic results showed that the 33-frame sequence clearly captured tracer arrival, early vascular signal, and gradual accumulation in brain tissue over time. Extracted time activity curves from representative cortical gray matter, deep gray matter, and white matter ROIs were stable and exhibited physiologically plausible regional ordering.
Conclusion
These initial human results demonstrate that the SmartBrain wearable PET system can support both static and dynamic FDG brain imaging. In static imaging, the system showed good agreement with a clinical PET scanner in relative regional uptake pattern. In dynamic imaging, it captured tracer arrival and progressive tissue accumulation and enabled stable ROI-based time activity curve extraction. These findings support the feasibility of wearable human brain PET and provide a foundation for further work on quantitative corrections and future behavioral and clinical applications.Speaker: Qiyu Peng (Shenzhen Bay Laboratory) -
28
Development and Evaluation of a High-Performance DOI-Enabled Dedicated Brain PET/CT System
Dedicated brain positron emission tomography (PET) systems have the potential to provide improved spatial resolution and sensitivity compared with conventional whole-body PET scanners. In this study, we developed and evaluated a depth-of-interaction (DOI)-enabled dedicated brain PET system designed for high-resolution brain imaging.
The detector modules employ LYSO crystal arrays (2.0 × 2.0 × 20.0 mm³) coupled to multi-pixel photon counter (MPPC) arrays. A light-sharing window (LSW) structure is implemented to estimate DOI and mitigate parallax errors. The system consists of ten detector panels with an axial field of view (AFOV) of 280.6 mm and a detector face-to-face diameter of 366.1 mm. System performance was evaluated according to the NEMA NU 2-2018 standard, including spatial resolution, sensitivity, and time-of-flight (TOF) resolution.
Experimental results show a mean spatial resolution of 2.25 mm. The measured total sensitivities are 16.12 cps/kBq at the center of the field of view and 16.75 cps/kBq at a radial offset of 10 cm. The coincidence timing resolution is 284 ps. The use of DOI encoding effectively compensates for parallax errors and helps maintain spatial resolution across the field of view. Preliminary brain imaging experiments further demonstrate that the proposed system can reveal fine anatomical structures in several brain regions.
These results indicate that the developed system provides high-resolution brain imaging capability with effective DOI-based parallax error compensation and good TOF timing performance, supporting its feasibility for dedicated high-resolution brain PET imaging.
Speaker: Mr Xin Yu
-
26
-
Coffee Break
-
Session 8: FTMI-SystemsConvener: Kuangyu Shi
-
29
Time-of-Flight Artefact Reduction in Limited-Angle PET: How Effective is it in Practice?
Limited-angle positron emission tomography (PET) is attracting increasing interest for a variety of applications, such as organ-specific PET, in-vivo verification in particle therapy, and cost-effective total-body PET (TB-PET). Various simulation studies have indicated that the artefacts that arise in limited-angle PET images due to missing lines-of-response (LOR) can be mitigated by means of time-of-flight (TOF) reconstruction, provided that the system coincidence resolving time (CRT) is about 200 ps FWHM. Here, we put this hypothesis to an experimental test for the first time. We investigate how the loss of angular information affects PET images of a Derenzo-like phantom, for various levels of angular coverage, by successively removing more and more detector modules from a full-ring monolithic-scintillator PET setup. This is done with the phantom placed isocentrically and at 20 cm radial offset. The detectors have a 1.7 mm full-width-at-half-maximum (FWHM) spatial resolution, a 212 ps FWHM CRT, and 4.7 mm FWHM depth-of-interaction (DOI) resolution. The images reconstructed from these experimental datasets are analyzed and several metrics quantifying the degree of image distortion are calculated, viz. the structural similarity index measure (SSIM), the normalized root mean square deviation (NRMSD), the signal-to-noise ratio (SNR), the eccentricity, and the peak-to-valley ratio (PVR) of the hot rods in the phantom. It is found that a partial ring with an angular coverage of two-thirds yields images that approach the quality of those of a full ring if both TOF and DOI are used in the reconstruction. However, complete removal of limited-angle artefacts is not yet achieved at a CRT of 200 ps FWHM. Moreover, the beneficial effect of DOI is found to be at least as important as that of TOF if the phantom is placed at 20 cm radial offset.
Speaker: Dr Dennis Schaart -
30
Improvements in Compton camera prototypes for targeted radionuclide therapy
Compton cameras show promise in medical applications in which the performance of gamma cameras is not optimal. This is the case of Targeted Radionuclide Therapy (TRT) assessment, which is gaining interest as a cancer treatment option. In this treatment modality, secondary photon emission from the therapeutic radionuclides can be employed to image the radiopharmaceutical distribution in the body. However, high photon energies and, in the case of alpha therapy, low activities, make conventional gamma cameras sub-optimal for this task. Compton cameras can overcome such problems offering better performance and higher image quality in TRT assessment.
The IRIS group has developed a Compton camera prototype made of monolithic LaBr3 crystals coupled to SiPM arrays that is being evaluated for secondary photon imaging in TRT. The detectors are read out with the ASIC VATA64HDR16 and operated with the AliVATA readout system. The latest prototype version, MACACO III+, is made of two planes. The first one consists of a single detector and the second one is made of four detectors. The prototype has been tested with custom-made Derenzo-like phantoms filled with F-18 and I-131 and also with irregular phantoms with hot spots in a warm background. In addition, mouse phantoms and living mice injected with I-131NaI have been successfully imaged. Simulations of mice undergoing treatments with Ac-225 show very promising results and experimental measurements with this radionuclide are planned.
Good timing resolution is a relevant parameter in Compton imaging, since the operation of the two planes in time coincidence makes it essential to reduce background. In addition, achieving a timing resolution better than 200 ps would allow to identify the backscattering events, further reducing background. Timing information can also be employed for a better determination of the photon interaction.
In spite of the excellent results achieved with the prototype MACACO III+, timing resolution remains to be a parameter to optimize in the prototype. Contrary to PET, Compton cameras have to deal with time walk in signal processing, degrading the timing resolution. In addition, timing resolution in monolithic detectors is challenging due to the spread of the light among the different photodetector pixels, resulting in strong light variations among channels. The group achieved a coincidence time resolution of 1.5 ns in Compton camera mode with the TOFPET2 ASIC from PETsys. However, nonlinearities in the electronics response resulted in a sub-optimal imaging performance.
Current attempts in the system performance improvement are twofold. On one hand, through the use of the FASTIC+ ASIC. On the other hand, the IRIS group coordinates the European project AIDER, which aims at the development of a high performance Compton camera for TRT. The developments include dedicated readout electronics for improved timing resolution.
Speaker: Prof. Gabriela Llosá (IFIC, CSIC-UV) -
31
Prototype Development of an On-Chip PET System with Dual-Sided Crystal Readout for Enhanced Gamma-Ray Interaction Point Localization
Background. Organ-on-chip (OOC) platforms mimic human tissue at millimeter scale and provide unique models for drug development and disease research. Quantitative imaging of radiotracer distribution within these devices can totally change their scientific utility, however no PET system exists at the physical scale and resolution required. We present a dedicated on-chip PET detector concept. It revolves around a new dual-sided readout architecture, whereby SiPM arrays placed on one large face and on the side face of the same monolithic LYSO crystal yield complementary spatial information, drastically enhancing scintillation location determination and image reconstruction quality.
Methods. The physical prototype features two monolithic LYSO crystals (5×5 cm² in area, 15 mm thick) mounted face-to-face, with the OOC unit in the space between them. Dual-sided readout is carried out by each crystal: the large 5×5 cm² face on one side is equipped with a 16×16 array of 256 SiPM elements, while the 15 mm side face is covered by two 8×8 SiPM matrices, with channels not having direct crystal coverage being excluded from the analysis chain. Light patterns across the entire SiPM geometry were generated by GATE Monte Carlo simulations to train a Convolutional Neural Network (CNN) for predicting gamma-ray interaction points. The consistent part of large-face-only readout versus combined large-face and side-face readout was measured, thus directly quantifying the supplementary localization information encoded by orthogonal side-face matrices. Then, the simulation-trained CNN was used on experimental data from the physical prototype to confirm the generalization from simulated to real detector data. Spatial resolution was further restored by applying a deep learning positron-range correction algorithm based on U-Net architecture.
Results. The dual-sided readout design clearly brought a significant enhancement over the large-face-only layouts: side-face SiPM matrices provided depth-of-interaction information which could not be obtained from the large face only, thereby reducing the positioning error. In the optimized setup, the mean positioning error was 0.80 mm, the system sensitivity was 34.81%, and the mean spatial resolution of the reconstructed image was 0.55 mm FWHM according to the simulation. The positron-range correction using deep learning significantly enhanced spatial resolution by 32%, thus it recovered over 91% of the maximum theoretically achievable gain. Most importantly, the simulation-trained network when applied to the real experimental data still showed reliable positioning results which implies that the Monte Carlo training framework generalizes to the physical detector measurements without the need for retraining.
Conclusion. The implementation of a two-crystal face-to-face configuration with dual-sided readout a 16×16 large-face panel combined with 8×8 side-face arrays for each crystal is the main hardware revolution of this system. Light distributions recorded at the same time from perpendicular crystal faces represent spatial coordinates in a complementary way, thus sub-millimetre reconstruction can be achieved without pixelated detectors. Establishing the workability of the system, the transport of simulation-trained AI models to real experimental data has been attested as effective. This detector design, merged with AI-based reconstruction, is capable of yielding the quantitative and spatial results necessary for the insertion of functional PET imaging directly with organ-on-chip devices, which allows radiotracer-based measurement of metabolic and pharmacokinetic processes in living microfluidic tissue models.
Speaker: Narendra Rathod (Inselspital and University of Bern) -
32
The ADMIRAL experiment: Ag-111 for imaging and therapy
The ISOLPHARM (ISOL technique for radioPHARMaceuticals) project, led by the Legnaro National Laboratories of the National Institute of Nuclear Physics, aims to produce a wide range of high-purity radioisotopes for medical applications, both diagnostic and therapeutic. A central role is played by SPES (Selective Production of Exotic Species), a second-generation ISOL facility that enables the production of innovative radionuclides with high purity and strong clinical relevance. Within this framework, the ADMIRAL experiment (2023–2025) focuses on the study of radiopharmaceuticals labeled with the emerging radionuclide Ag-111, with the goal of evaluating its diagnostic and therapeutic potential.
Ag-111 exhibits several properties that make it particularly attractive for nuclear medicine: a suitable half-life, medium-energy beta emissions useful for therapy, and medium-energy gamma emissions that can be effectively detected using planar gamma cameras or SPECT devices. These characteristics make it a promising candidate for theranostic applications, where diagnosis and therapy are combined within the same radiopharmaceutical.
The experimental activities covered the entire development chain, from radionuclide production to the assessment of biological effects. Ag-111 was produced via neutron-gamma reactions starting from an enriched palladium target (Pd-110) at the TRIGA Mark II nuclear reactor of the Laboratory of Applied Nuclear Energy (LENA) in Pavia. Particular attention was devoted to optimizing radiochemical procedures, including the dissolution of the irradiated target and the purification of the isotope, in order to achieve high radionuclidic purity. Once produced, the radionuclide was incorporated into macromolecular structures designed to selectively transport it to tumor tissues, promoting specific interaction with cancer cells.
In parallel, a novel system for high-resolution two-dimensional β-imaging was developed. This detector is based on ALPIDE sensors, originally designed for the inner tracking system of the ALICE experiment, and exploits Monolithic Active Pixel Sensors (MAPS) technology. A dedicated mechanical support enabled the acquisition of high-resolution β-images of both conventional 2D cell cultures and thin three-dimensional structures (2.5D scaffolds), allowing detailed investigation of radiopharmaceutical distribution at the microscopic level.
Complementary to this, a gamma imaging module was designed and constructed to detect the γ radiation emitted during Ag-111 decay. The work focused on optimizing the collimator geometry, improving the coupling between scintillators and silicon photomultipliers (SiPMs), and developing the associated electronics and data acquisition system. The resulting planar imaging system enabled effective detection of Ag-111 and allowed comparison with existing preclinical imaging devices.
Finally, the biological effects of Ag-111-based targeted radionuclide therapy were investigated through a dedicated radiobiological approach. A series of in vitro experiments were performed, including clonogenic assays, DNA damage analysis, and cellular uptake measurements. These results were correlated with the absorbed dose at the cellular level, calculated using Monte Carlo simulations with the Geant4 toolkit and its Geant4-DNA extension, which allows simulations at microscopic scales.
This work aims to summarize the objectives achieved by the ADMIRAL experiment, illustrating the main studies conducted and the key results obtained.
Speakers: Alberto Arzenton (University of Padua), Aurora Leso (University of Ferrara), Davide Serafini (INFN-LNL), EMILIO MARIOTTI (Università di Siena), Giulia Saveria Valli (University of Siena), Marcello Lunardon (University of Padova), Stefano Corradetti, alberto andrighetto (INFN-LNL)
-
29
-
-
-
Session 9: PSMR-Overview & DetectorsConvener: Roger Lecomte
-
33
PET & PET/MR in 2026: Toward Fully Integrated, Quantitative, and Dynamic Multimodal Imaging
PSMR introductory talk
Speaker: Nicola Belcari -
34
Analytical modeling of inter-crystal scattering in PET with energy information
Including inter-crystal scattering (ICS) events in Positron Emission Tomography (PET) can be beneficial in low-count scenarios such as small-animal imaging, where sensitivity is a limiting factor. To this end, conventional ICS-recovery techniques typically aim to identify the original line of response, usually with insufficient success rates. We have recently proposed a new physics-based analytical model for ICS events that obviates the aforementioned need. There, the additional information provided by the measured deposited energies in ICS events was not used. In this work, we investigate how this information can be integrated in the system matrix by using an energy-uncertainty model. Monte-Carlo-simulated data of the MERMAID small-animal PET was used to assess the effects on an image quality phantom. Ongoing results show that, while using the detected deposited energies might be beneficial to improve the spatial resolution when using ICS events, these benefits are constrained to the detector energy and spatial resolution.
Speaker: Dr Jorge Roser (Institute of Medical Engineering, Universität zu Lübeck) -
35
Accurate 3D Positioning in Semi-Monolithic PET Detectors including ICS recovery
PET image quality depends on accurately determining the lines of response, which requires precise 3D localization of annihilation photon interactions within the detectors. In pixelated crystal arrays, spatial resolution is inherently limited by crystal size and can only be improved by increasing granularity, at significant cost. Monolithic and semi-monolithic arrays overcome this by inferring interaction position from the photosensor light distribution, enabling high spatial accuracy through calibration and software alone.
We present the full 3D position calibration of the Ultra-High-performance Brain (UHB) PET scanner, based on semi-monolithic detectors, using a collimated fan beam and multilayer perceptron (MLP) networks. Overall resolutions of 2.3$\pm$0.6 mm and 2.7$\pm$0.8 mm FWHM were achieved along the monolithic and depth-of-interaction directions, with individual mini-modules reaching ~1.6 mm and ~2.0 mm FWHM, respectively. Slab identification accuracy, applying ICS recovery, reached 78%.
These results demonstrate that semi-monolithic detectors can deliver high 3D spatial resolution across all directions, supporting their application in next-generation brain PET scanners.
Speaker: Dr Fiammetta Pagano (Instituto de Instrumentación para Imagen Molecular (i3M) – CSIC) -
36
Multiple Emission Tomography: A High-Sensitivity Simultaneous PET and Compton-Camera scanner
The growing importance of theranostics and research with long‑lived isotopes demands imaging systems far more sensitive than standard PET or SPECT. Many emerging isotopes lie outside current detector capabilities, resulting in low sensitivity, high background, and poor quantification, as seen with $^{90}$Y verification and $^{124}$I or $^{89}$Sr imaging. The Multiple Emission Tomography (MET) project introduces a novel imaging device based on inorganic scintillators, capable of operating in dual modes: PET-style coincidence detection of 511 keV photon pairs and single-photon Compton-scattering detection in a single-layer configuration. The technical configuration of the MET prototype utilizes a DOI-capable light-sharing scheme. Each module consists of a 16 $\times$ 16 matrix of depolished LYSO crystals coupled to an 8 $\times$ 8 array of Hamamatsu Silicon Photomultipliers (SiPMs). This design, supported by a custom algorithm to solve Compton kinematics, allows for the precise measurement of the photon’s depth of interaction (DOI) and the identification of the first crystal of interaction in scattering events. Latest results from the characterization of the first MET prototype modules show promising performance. The detector has achieved an average energy resolution of 12.1 $\pm$ 0.1 FWHM at 511 keV and an average DOI resolution of 3.5 $\pm$ 0.1 mm FWHM. While current results use the TOFPET2 ASIC, the project is transitioning to the ALCOR ASIC, which will offer higher bandwidth (up to 2 Mevents/s per channel) and 25 ps time binning to support high-rate Compton reconstruction. This technology is expected to significantly impact theranostics, collimator-less SPECT, and in-vivo range verification for Charged Particle Therapy.
Speaker: Marco Pizzichemi (Universita Milano-Bicocca (IT) and CERN) -
37
Investigation of SiPM Channel Compression Schemes
A prevailing trend in PET development is the use
of smaller SiPMs to achieve higher spatial resolution. However,
the resulting increase in SiPM quantity places a heavier burden
on the processing hardware. To address this without requiring
upgrades to the underlying readout electronics, multiple SiPM
signals can be multiplexed into a single channel. This approach is
utilized in the Hyperion PET detector platform, which integrates
both SiPMs and digitizing ASICs. The platform consists of sensor
tiles connected to the module mainboards via flexible cables.
By leveraging versatile analog SiPM and ASIC combinations,
these tiles are adaptable for various PET/MRI projects. The
initial implementation will feature a 48 x 48 mm2 form factor
utilizing 4 mm SiPMs. Through signal compression, the 144
SiPMs will be read by a single 64-channel TOFPET ASIC,
significantly optimizing spatial footprint, power consumption,
and overall costs. A range of channel compression schemes
with a factor of at least 2.25 are currently under investigation.
Five different compression schemes that do not use additional
electronic components are proposed. For comparison the row-
column-sum approach with additional resistors was also included
in the investigations. For local compression schemes SiPMs in
close proximity are connected in parallel while for more global
compression schemes distances between connected SiPMs are
maximized. The simple patterns allow straightforward algorithms
for crystal identification, whereas other more optimized patterns
employ more advanced processing algorithms, like neural net-
works, for this task. Measurements with 4 mm pixels show that
the mean absolute error only slightly increases from ∼1.25 mm
to ∼1.32 mm when using one of the local compression schemes in
comparison to using uncompressed data. Furthermore measure-
ments with 2 mm Slabs show using smaller 4 mm SiPMs with
compression schemes was still advantageous over using bigger
6 mm SiPMs for almost all investigated compression schemes. The
energy resolution is also effected by the choice of compression
scheme. Preliminary results show that using local compression
schemes as tetris and fishbone can even improve the energy
resolution over non compressed data. Test PCBs were designed
for SiPMs and for the compression schemes as interposing
PCBs. At the same time, a sensor tile is being designed with
a TOFPET2 ASIC and an FPGA on the electronics side and just
two connectors for the SIPMs on the other side. Consequently,
the digitizing ASIC, compression schemes, and SiPMs can be
combined freely.Speaker: Thore Meyer (Chair of Imaging & Computer Vision, RWTH Aachen University)
-
33
-
Coffee Break
-
Session 10: PSMR-SystemsConvener: Nicola Belcari
-
38
Hybrid PET-MRI-FUS: Performance Validation, Trimodal Imaging, and 9.4T Positron Range Confinement
Hybrid systems integrating Positron Emission Tomography (PET), Magnetic Resonance Imaging (MRI), and Focused Ultrasound (FUS) are highly attractive for preclinical imaging and therapy monitoring, yet no commercial trimodal solution is currently available. In this work, we present the development and validation of a dedicated preclinical PET insert designed for simultaneous operation with high-field MRI and commercial FUS devices, and we demonstrate its potential in both trimodal imaging applications and positron-range confinement studies at 9.4 T. The system is based on monolithic LYSO crystals and provides an axial field of view of 67 mm. Performance evaluation, inspired by the NEMA NU-4 2008 protocol, showed a homogeneous submillimeter spatial resolution of 0.9 mm with depth-of-interaction capability, a sensitivity of 3.8%, and a peak Noise Equivalent Count Rate of 80 kcps. Image quality measurements yielded recovery coefficients up to 0.89 and spill-over ratios of 11% in air and 22% in water, demonstrating performance comparable to state-of-the-art preclinical PET systems.
Beyond technical validation, the PET insert was employed in different trimodal experiments. In a first phantom proof-of-concept, the system was combined with a custom low-field MRI and a dedicated FUS device. Focused ultrasound was used to locally heat the phantom and melt a gelatin barrier, enabling the redistribution of an initially confined ¹⁸F-FDG solution. This process was successfully monitored over time by PET-MRI, showing the ability of the platform to track FUS-induced changes in tracer distribution. In addition, in vivo feasibility was demonstrated in murine brain studies using the PET insert together with a 9.4 T MRI scanner and a commercial FUS system. Blood–brain barrier opening was induced with microbubbles, and co-administered Gd-DOTA and ⁶⁴Cu-DOTA allowed MRI and PET to confirm co-localized enhancement in the sonicated regions.
Finally, the system was used to experimentally investigate positron-range confinement under strong magnetic fields. Thin glass capillaries filled with ¹⁸F, ⁸⁹Zr, and ⁶⁸Ga were imaged inside and outside a 9.4 T MRI scanner in different phantom materials and orientations relative to the magnetic field. While ¹⁸F showed minimal changes, higher-energy emitters exhibited a clear anisotropic reduction of positron-range blurring inside the MRI. For ⁸⁹Zr, improvements of about 25% were observed, whereas for ⁶⁸Ga the reconstructed FWHM decreased by up to 32–39%, with FWTM reductions reaching 56–58% depending on phantom density. These findings were further supported by microDerenzo measurements, where the magnetic field enabled improved rod separation for high-energy emitters, including resolution of 1.0 mm rods for ⁶⁸Ga at 9.4 T. Overall, this work demonstrates the versatility of the proposed trimodal PET-MRI-FUS platform for system validation, image-guided therapy monitoring, and high-field preclinical PET studies.
Speaker: Fernando Lopez Berenguer (Instituto de Instrumentación para Imagen Molecular (I3M), CSIC-UPV) -
39
First Results from the SAVANT Ultra-High-Resolution Phoswich-Based DOI Brain PET Scanner
High spatial resolution is essential to successfully resolve small structures in the brain and to enable imaging of neurological pathways at the onset of diseases. A key challenge with ultra-high-resolution PET scanners is the significant degradation of resolution that occurs beyond the central area of the field-of-view (FOV). The Ultra-High-Resolution (UHR) Brain PET scanner was designed with short (12 mm) LYSO crystals to mitigate the parallax error off the FOV center, but at the expense of reduced sensitivity [1]. To address these limitations, the LabPET II based technology platform of the UHR was upgraded to implement depth-of-interaction (DOI) measurement using longer (15 mm) dual LGSO phoswich detectors in the Scanner Approaching in Vivo Autoradiographic Neuro Tomography (SAVANT). Simulation results indicate an estimated gain in sensitivity of 40%, while maintaining the resolution below 2 mm FWHM at the edge of the brain. For DOI determination, the crystal identification between slow (43-48 ns decay time) and fast (30-35 ns decay time) LGSO scintillators uses a model-based dual-threshold time-over-threshold (ToT) discrimination technique that is currently achieving better than 70% accuracy. The 6.5/8.5 mm length ratio of the top slow and bottom fast LGSO scintillators resulted from trade-offs between off-center spatial resolution degradation and balanced coincidence detection efficiency of the two crystal layers. We report the first experimental results obtained using the partially assembled scanner (182 mm axial extent), which demonstrate the improvement in resolution uniformity across the radial FOV achieved using this simple DOI discrimination scheme. Point source and phantom measurements taken at various radial positions within the FOV effectively demonstrate the gain in spatial resolution compared to non-DOI measurements, though not reaching the resolution predicted by simulation. It is anticipated that further refinement of the crystal identification process, potentially using AI, will yield the forecasted benefits from simulation. Preliminary results for imaging the human brain will be presented.
[1] E. Gaudin, M. Toussaint, C. Thibaudeau, M. Paille, R. Fontaine, and R. Lecomte, “Performance Simulation of an Ultrahigh Resolution Brain PET Scanner Using 1.2-mm Pixel Detectors,” IEEE Trans. Radiat. Plasma Med. Sci., vol. 3, no. 3, pp. 334–342, May 2019, doi: 10.1109/TRPMS.2018.2877511.
Speaker: Prof. Roger Lecomte (Université de Sherbrooke, Sherbrooke Molecular Imaging Center of CRCHUS, IR&T Inc.) -
40
Simulation-Based Performance Evaluation of the NeuroSphere PET Insert for 7T-MRI
Introduction The Human Dynamic NeuroChemical Connectome (HDNCC) scanner is a novel, brain-dedicated imaging system that integrates the high spatio-temporal resolution NeuroSphere PET insert with a Siemens MAGNETOM Terra.X 7T MRI scanner. While PET-MR technology offers exceptional molecular sensitivity, existing systems often lack the temporal resolution required to capture dynamic changes on time scales comparable to cognitive processes—a capability typically reserved for functional MRI. To address this, the NeuroSphere leverages a unique spherical geometry and state-of-the-art detectors to significantly boost sensitivity and spatial resolution. This work presents a comprehensive simulation of the NeuroSphere’s performance following NEMA protocols, demonstrating its potential to achieve functional time frames on the order of seconds.
Materials and Methods The NeuroSphere is composed of 872 detector modules arranged in a spherical configuration with a 32 cm inner diameter. This design provides a solid angle coverage of approximately 70% around the human head, which is critical for enhancing signal-to-noise ratio in dynamic studies. Each detector block consists of a 10x10 array of 1.6x1.6x26 mm3 LSO crystals coupled to a 4x4 matrix of MPPCs (4x4 mm2). Light sharing for Depth-of-Interaction (DOI) identification is enabled by a 380 μm methacrylate light guide applied to the entry face of each block, achieving a DOI resolution of 7 mm.
System performance was evaluated using the GPU-based UMC-PET Monte Carlo simulator, which offers high-fidelity voxelized modeling and operates up to 2,000 times faster than CPU-based alternatives. Detector coordinates were imported directly from the CAD model to create a realistic digital twin, including realistic detector modeling based on actual characterization measurements indicating an energy resolution below 13% and a coincidence timing resolution of 526 ps. Simulations followed NEMA NU-2-2018 standards to estimate spatial resolution, count rates, and sensitivity. Sensitivity was further optimized by exploiting triple coincidences from inter-detector scatter events. Image reconstruction was performed using a 3D-OSEM algorithm incorporating spatially variant point-spread-function (SV-PSF) modeling to ensure high resolution across the entire field of view (FOV).
Results Simulations demonstrate an outstanding average sensitivity over 30% for a 435-585 keV energy window across the brain. The system achieves a homogeneous spatial resolution of better than 1.6 mm; reconstructed images of a Derenzo phantom confirmed that 1.6 mm rods are clearly identifiable at both centered and 80 mm off-center positions.
Regarding count rate performance, the peak Noise Equivalent Count (NEC) rate was obtained over 20 kBq/mL (440 MBq in a 200 mm diameter and 700 mm length phantom). For typical brain activity levels, the system maintains high trues-to-prompts ratios, with approximately 500 kcps trues 100 MBq total activity (<5 kBq/mL).
Conclusion The NeuroSphere PET insert shows exceptional potential for high-speed dynamic brain imaging. Its high sensitivity and sub-1.5 mm spatial resolution offer the capability to push PET temporal resolution toward levels comparable to fMRI. The first complete assembly of the NeuroSphere is currently undergoing experimental calibration, which will further refine these performance estimates for future clinical and research applications.Speaker: PABLO GALVE LAHOZ (Institute for Physical and Information Technologies “Leonardo Torres Quevedo”, ITEFI, Spanish National Research Council (CSIC), Madrid, Spain) -
41
Experimental Results of a Brain PET Insert inside a 3T Siemens Prisma MRI
Brain Positron Emission Tomography (PET) imaging provides functional assessment of different physiological and pathophysiological processes in the human brain. It can therefore be used to evaluate tumors, infections, inflammation, and various neurodegenerative diseases, such as Alzheimer or Parkinson, among others. The combination of PET with Magnetic Resonance Imaging (MRI) enhances our understanding of the mechanisms underlying normal brain function and neurological disorders.
In this work, we have constructed and evaluated a new PET insert for brain imaging, named DeepBrain. The system is based on curved monolithic scintillation crystals and an edgeless geometry. The 2D annihilation impact coordinates were estimated using neural networks, and the energy and depth of interaction values were calibrated using a Voronoi-based approach. We report the system’s performance and its compatibility with a 3T MRI system.
An average spatial resolution of 1.9 ± 0.1 mm, 2.0 ± 0.1 mm, and 1.5 ± 0.1 mm FWHM was achieved at the center of the axial Field of View (FOV) for the radial, tangential, and axial directions, respectively, after correcting for parallax errors. A maximum sensitivity value of 1.5% was measured at the center of the FOV. The noise equivalent count rate peak reached 43 kcps at 1.7 mCi. Small differences were measured in the percent contrast and background variabilities for a dedicated image quality phantom when running a variety of pulse sequences in the 3T MRI. The next steps are implementing scatter and quantification corrections and the development of a dedicated RF coil.Speaker: Marta Freire Lopez-Fando (Instituto de Instrumentacion para Imagen Molecular (i3M)) -
42
Characterization of an Intravital Theranostic Imaging Camera
Background. Spatial heterogeneity of radiotracer distribution within the tumour microenvironment is a primary determinant of radiopharmaceutical therapy (RPT) efficacy and normal-tissue toxicity. Quantifying this heterogeneity in living tissue at the mesoscale the biologically relevant length scale of 50–500 μm that encompasses individual micro-vessels, hypoxic gradients, and receptor-expression boundaries remains inaccessible to current measurement methods. Clinical PET and SPECT provide in vivo compatibility but are limited to millimetre-scale spatial resolution; digital autoradiography offers sub-100 μm resolution in fixed thin sections but requires tissue sacrifice, hours-long exposures incompatible with kinetic measurement, and suffers substantial resolution degradation at physiologically relevant specimen thicknesses. The Intravital Theranostic Imaging Camera (ITIC) is a single-particle ionizing-radiation quantum imaging system developed to address this methodological gap.
Methods. ITIC detects individual beta particles through event-by-event quantum acquisition: each particle interaction in a thin scintillation screen generates a spatially localised flash of light that is amplified by a dual-stage microchannel plate (MCP) image intensifier and recorded by a scientific CMOS sensor as a list-mode entry encoding position, timestamp, and energy. System performance was comprehensively characterised by multi-modal comparison against gamma counting and phosphor-plate autoradiography using ¹⁸F calibration standards and heterogeneous biological leaf phantoms with activity contrasts mimicking the tumour and its microenvironment. Spatial resolution was independently validated against optical microscopy ground truth by two complementary analysis frameworks. A GEANT4 Monte Carlo simulation established the empirical FWHM-versus-tissue-thickness transfer function across the full intravital window chamber thickness range. Temporal and dosimetric performance was assessed through longitudinal ¹⁷⁷Lu-PSMA imaging of ex vivo human prostate cancer xenograft tissue over six days.
Results. ITIC achieved quantitative accuracy comparable to gamma counting, with systematic errors under 5% in the entire activity range. In addition, spatial resolution of 115 ± 31 μm FWHM was achieved in biological phantoms mimicking tissue thicknesses relevant to intravital imaging, with the whole image acquisition taking 15-30 minutes, as opposed to 4-12 hours required for autoradiography. More importantly, while the resolution of autoradiography falls off to 150-250 μm in specimens thicker than 100 μm, the resolution of ITIC remains the same even in tissue phantoms of 200-400 μm, which are directly compatible with the dorsal window chamber geometries that are used for intravital imaging. The Monte Carlo-derived FWHM-versus-tissue-thickness transfer function, the first empirical model of its kind for beta-particle imaging through soft tissue, quantified resolution across three physically distinct regimes and provides a principled basis for intravital experimental design at any window chamber thickness. Longitudinal 177Lu imaging showed spatial heterogeneity patterns that were stable (tissue contour plot) while the decay kinetics were in line with the physical half-life, thereby confirming the count-rate measurement pipeline as the quantitative baseline for future dose-point-kernel convolution to absorbed dose maps. As an initial sign of more dosimetric capabilities, time-resolved count-rate curves retrieved from spatially different tissue compartments, high-uptake tumour and low-uptake parenchyma, displayed physically consistent independent decay behaviour, implying that ITIC has the spatial specificity and temporal resolution to enable spatially resolved microdosimetry in future dedicated studies.
Conclusion. ITIC is an imaging system that can perform quantitative, real-time radiopharmaceutical imaging of living tissue at the mesoscale, the spatial scale at which the therapeutic outcome is determined. This characterization is the quantitative basis for window chamber studies, which allow direct, longitudinal observation of radiotracer distribution, uptake kinetics, and dose heterogeneity within tumour and its microenvironment in a living organism. These are measurements that require the simultaneous combination of spatial resolution, temporal continuity, and biological compatibility that ITIC uniquely brings together as an integrated intravital platform.
Speaker: Narendra Rathod (Inselspital and University of Bern)
-
38
-
Lunch Break
-
Session 11: PSMR-MethodsConvener: Giancarlo Sportelli
-
43
Breast PET Insert in a Whole-Body PET/MRI: Attenuation Effects and Resolution Recovery
We are developing a breast PET insert operating in combination with a whole-body (WB) PET/MRI scanner to improve the detectability of small lesions and early metastases. However, additional detector material can introduce attenuation-induced artifacts that degrade WB-PET image quality. We implemented a joint reconstruction with attenuation correction to explicitly handle the different types of coincidence from the insert and the WB-PET and thereby improve the spatial resolution. To quantify the effectiveness of our method, Monte Carlo simulations were performed using $^{22}$Na point sources at different positions to assess spatial resolution, and a hollow-sphere phantom to evaluate image quality and contrast. Insert attenuation reduced detected WB-PET coincidences by 23.6% at the central field-of-view and degraded spatial resolution of WB-PET-only events by up to 3.0 mm $\Delta$FWHM. Nevertheless, the joint reconstruction including coincidences between the insert and the WB-PET recovered spatial resolution, yielding FWHM values of 3.05 mm (x), 2.34 mm (y) and 2.60 (z) mm compared to 7.17 mm (x), 6.08 mm (y), 3.60 mm (z) for the WB-PET-only events. The inclusion of all coincidence types improved the visibility of the sphere inserts in the phantom and increased contrast recovery for the 8 mm sphere from 28% to 33%. Future work will focus on comprehensive image quality assessment, including space-invariant point spread function modeling, scatter correction, and inter-crystal scattering model implementation, followed by evaluation with realistic anthropomorphic phantoms.
Funding: The project is supported by the German Research Foundation (DFG) under grant agreement no. 508064995, and high-performance computer "Lise" at the NHR@ZIB, project no. shb00004.Speaker: Hong Phuc Vo (Institute of Medical Engineering, Universität zu Lübeck) -
44
A long axial field-of-view PET insert design based on dual-ended side-readout for simultaneous PET and metabolic MRI in the META-scan system
Introduction:
The META-scan, an ultra-high-field Magnetic Resonance Imaging (MRI) scanner developed to visualize metabolic processes in vivo in the entire human body, can potentially be used to study acute metabolic effects and fast interactions between the brain and other organs, such as the liver, gut, heart, and kidneys. However, due to its low sensitivity, studies involving drug tracing require unacceptable pharmacological doses. In addition, simultaneous imaging of fast direct metabolic interactions between organs more than 40 cm apart is currently not possible due to the limited field-of-view. We propose to address these two limitations by developing a positron emission tomography (PET) insert for the META-scan, to extend its metabolic MRI capabilities with the possibility to perform simultaneous PET for drug tracing and/or metabolic imaging anywhere in the body. Since the insert is aimed at enabling dynamic imaging of the fast metabolic processes for which simultaneous imaging is essential, the sensitivity of a long axial FOV (LAFOV) PET system design is needed. However, incorporating a LAFOV design into the META-scan presents a major challenge, since the available physical space is severely limited due to the relatively small diameter of its 7T bore coil. The aims of this study were to explore the possibilities of employing inorganic crystal fiber arrays with dual-ended side-readout to create a LAFOV detector geometry for a PET insert that fits within the physical space constraints of the META-scan, to give an initial estimation of the expected sensitivity and dynamic performance of the various options, and to compare them to the expected sensitivity of an optimized more traditional design based on back-readout of the crystals.Methods:
For both design strategies (back-readout and side-readout), we modelled a range of system designs using GATE with varying axial FOV lengths, numbers of crystal arrays, crystal thicknesses, and axial gap sizes, while accounting for the physical space required for the crystals and the data readout and power supply cables. As a rough estimate of the expected sensitivity of each design, we simulated a 4-minute acquisition of a human body-mimicking phantom containing 33.55 MBq F-18-FDG and expressed the number of true coincidences as a percentage of the number of true coincidences recorded from the phantom in a GATE model of the uEXPLORER system.Results:
Even when modelling only crystals and cables, assuming zero thickness for crystal readout electronics and cooling system, the side-readout designs achieved better sensitivity than the most optimal back-readout designs for axial FOVs longer than 1.5 m. Our preliminary side-readout design with 192 cm long axial FOV recorded 35.4% of the number of true counts recorded by the uEXPLORER system, suggesting such a system could acquire a total-body PET image of diagnostic quality in a time span in the order of 2 minutes and a temporal resolution of ~0.3 s.Conclusion:
Using dual-ended side-readout, we expect to be able to develop a total-body PET insert for the META-scan system, with sufficient axial FOV length and sensitivity to perform simultaneous drug tracing or metabolic PET imaging anywhere in the human body.Speaker: Woutjan Branderhorst (University Medical Centre Utrecht) -
45
3D-printed multimodality phantoms for quality control procedures in TBPET, PSMR and FTMI systems.
Background:
Total Body PET (TB-PET) and PET/SPECT combined with MR (PSMR) systems, and Fast Timing in Medical Imaging (FTMI) have changed the field of molecular imaging. The technical improvements conveyed by these next-generation systems offer a unique opportunity to develop new scanning protocols, enabling earlier diagnosis and more precise, low-dose whole-body imaging through dramatically improved sensitivity, multimodal integration, and temporal resolution. However, their translation into clinical practice requires the development of advanced, dedicated phantoms and quality control (QC) protocols capable of addressing key performance limitations, including uniformity across extended fields-of-view (FOV), ultra-fast timing resolution, and multimodality compatibility without compromising quantitative and image quality accuracy.
Three-dimensional (3D) printing technology enables the fabrication of phantoms with anatomies and structural patterns precisely tailored to specific requirements, constituting a significant advancement for the optimization of QC procedures. This work proposes advanced QC approaches based on 3D-printed multimodality phantoms to quantitatively evaluate TB-PET, PSMR, and FTMI performance, with particular focus on extended FOV response, timing resolution, and cross-modality consistency.
Materials and methods:
A 3D-printing workflow was implemented. STL model optimization procedures were specifically tailored according to the phantom type, distinguishing between image segmentation in anatomical phantoms and pattern design for geometrical phantoms. Phantom fabrication combined Fused Filament Fabrication (FFF) and VAT-photopolymerization techniques.
Phantom imaging was performed in GE SIGNA PET/MR with 3 T magnetic field, Philips Gemini TF64 PET/CT, Philips Vereos PET/CT and Siemens 3T TimTrio MR with a BrainPET insert to prove the QC procedure vaibility.
Results:
3D-printed geometrical phantom was fabricated for QC procedures on PET hybrid systems (Figure1.a), assesing co-registration, uniformity and distortion along 38 cm superior-inferior direction, resolution and distortion across a 28 cm FOV and concetration accuracy in the presence of human-tissue densities. Multimodality realistic brain, head-and-neck, thorax and prostate phantom was fabricated (Figure1.b), allowing more specific QC post-processing tools, such us motion compensation (respiratory for lung and bulk for brain) and segmentation models. Brain–cerebrovascular phantom (vessel diameters 2–5 mm) with and infusion–perfusion pump (10 to 80 ml/min) was fabricated for QC of PET dynamic protocolos (Figure 1.c). Preliminary results with the different imaging equipments proved the viability of the proposed QC protocols.
Conclusions:
The 3D-printed phantoms presented are available upon request and represent a valuable resource for developers of medical imaging systems and post-processing tools, enabling the quantitative assessment of image quality prior to clinical implementation.
Speaker: Luisa María Gavier Moreno (Grupo de Investigación Biomédica en Imagen, Instituto de Investigación Sanitaria La Fe) -
46
Multiple Emission Tomography: GPU-Accelerated LM-OSEM Reconstruction for Compton Events
The Multiple Emission Tomography (MET) detector is a single-layer ring of scintillator modules, that can work in Compton camera mode on top of the conventional "PET" one. Provided that a fast reconstruction that can exploit its high sensitivity can be developed, it is expected to enable imaging of exotic nuclei with high-energy gamma emissions. In this work, we present a list-mode Ordered Subsets Expectation Maximization (LM-OSEM) extension of the CoReSi Compton camera reconstruction adapted for MET. CoReSi was originally designed for standard two-layer Compton cameras and we extend it to operate on the single-layer geometry. We also replaced the native Maximum Likelihood Expectation Maximization (MLEM) solver with OSEM for faster convergence, and introduce event batching to decouple GPU memory usage from dataset size. Compunding all the improvements introduced enables the reconstruction of one million events in less than 5 minutes on consumer-grade hardware
Speaker: Matteo Neel Colombo (Universita & INFN, Milano-Bicocca (IT)) -
47
The MERMAID Project: Current Developments and Future Directions
The MERMAID (Multi-Emission Radioisotopes – Marine Animal Imaging Device) project aims to develop a dedicated PET/CT scanner for functional imaging of adult zebrafish. The system consists of four rotating detector modules arranged in pairs and provides a field of view (FOV) of 2.5 cm (diameter) and 6 mm (axial length). To extend the axial FOV, a motorized stage is integrated for bed placement and automatic bed positioning during the scan. Radioactive decay is compensated for during the measurement, and randoms are estimated using the singles-rate method. A dedicated list-mode MLEM reconstruction algorithm was developed in-house, including random and parallax compensation. 3D printing of radioactive resin was used to construct a downscaled NEMA NU-4-like image quality (IQ) phantom, used for IQ evaluation. The percentage standard deviation (uniformity) reached a local minimum of 10.65% at iteration 7. At iteration 20, the spill-over ratio (SOR) was 0.19 (SOR) while the recovery coefficient varied from 0.21 (rod diameter=1.0 mm) to 1.19 (rod diameter=2.5 mm). Additionally, in vivo scans of adult zebrafish were carried out, after intraperitoneal injection of 18F-FDG and a 30-minute incubation time. The fish were immobilized within a dedicated fish chamber with continuous freshwater and anesthesia flow. In the reconstructed images, several organs could be visualized, including the brain and swim bladder. Future work includes the integration of a CT scanner for attenuation correction and anatomical guidance, as well as the addition of four further PET modules to improve sensitivity and extend the axial field of view.
Speaker: Ms Rebecca Kantorek (Universität zu Lübeck)
-
43
-
Coffee Break
-
Session 12: PSMR-ApplicationsConvener: Casper Beijst
-
48
Sponsor Keynote: TERAPET:Translating Advanced Detection Technologies into Clinical Practice: The Terapet Portfolio
Platinum sponsor talk
Speaker: Christina Valgreen -
49
SPECT image reconstruction using PET priors via 3D Diffusion with Posterior Sampling for ultra-low count projections for alpha therapy dosimetry
Background: Alpha-particle radiopharmaceutical therapy (α-RPT) is a promising treatment modality for cancers such as metastatic castration-resistant prostate cancer. Compared to β-emitters (e.g. ¹⁷⁷Lu), α-emitters such as ²²⁵Ac deliver high linear energy transfer over a short range, enabling effective tumour cell killing while sparing surrounding tissue. However, clinical translation is limited by the lack of reliable quantitative imaging and dosimetry. SPECT imaging in α-RPT is particularly challenging due to low administered activities, resulting in highly ill-conditioned reconstruction problems with very limited photon counts. Conventional methods such as MLEM are highly sensitive to noise and regularized approaches are often insufficient while many AI-based methods often fail under such extreme conditions.
In this work, we propose a fully three-dimensional diffusion with posterior sampling (3D-DPS) framework for low-count SPECT reconstruction aiming to improve image quality while keeping data fidelity.
Methods: Our method leverages a generative diffusion model trained on 3D PSMA PET data to provide a population-based prior for radiotracer distribution. Unlike conventional DPS approaches based on l2 norm minimization, we formulate the data-consistency term using a Poisson log-likelihood consistent with standard iterative reconstruction methods for emission tomography such as MLEM/OSEM.
The diffusion model was trained using a variance-exploding stochastic differential equation with 1000 steps on 120 PSMA PET volumes, generating 3D activity distributions (128×128×128). During reconstruction, the learned prior is combined with a physics-based forward SPECT model through a data-consistency term.
Validation was performed using realistic SPECT simulations incorporating attenuation, collimator blur, and detector resolution. Five patient cases were simulated using 30 projection angles and ultra-low-count Poisson noise (~10⁵ detected photons). The proposed method was compared to MLEM and MLEM with total variation (TV) regularization.
Results and discussion: The proposed 3D-DPS significantly improves reconstruction quality compared to baseline methods. Quantitatively, we observed lower reconstruction error (NMAE = 0.36 ± 0.05) compared to MLEM (0.53 ± 0.02) and MLEM+TV (0.41 ± 0.05), corresponding to relative error reductions of ~32% and ~12%, respectively. Structural similarity is also improved (SSIM = 0.89 ± 0.05) compared to MLEM (0.77 ± 0.13) and MLEM+TV (0.85 ± 0.08), indicating better preservation of image features. Visually, the method suppresses noise while maintaining contrast and structural detail, whereas MLEM is dominated by noise and TV regularization leads to over smoothing and insufficient regularization for ultra low count scenarios.
Note that these improvements come with a preserved consistency with the measured data, with projection-domain log-likelihood values within 2% of standard MLEM reconstruction. This supports the assumption that the gains are not achieved at the expense of data fidelity, but rather through improved regularization driven by the learned prior.
Conclusions: In conclusion, the proposed framework provides a robust and physically consistent approach for reconstructing ultra-low-count SPECT data in α-RPT. This method has the potential to improve quantitative imaging and enable more reliable dosimetry in challenging low-count scenarios. Future work will focus on validation using Monte Carlo simulations and clinical datasets.Speaker: Alejandro López Montes (Inselspital Bern, University of Bern) -
50
Cross-Modal Transfer of MRI Foundation Model Embeddings Enables Multimodal FPIA PET–MRI Prediction of IDH Mutation in Glioma
Accurate non-invasive prediction of isocitrate dehydrogenase (IDH) mutation status remains a major challenge in glioma imaging. Although magnetic resonance imaging (MRI) provides valuable structural and functional information, conventional radiological features lack consistent predictive power across datasets. Recent advances in foundation models trained on large-scale MRI datasets offer a new opportunity to extract transferable imaging representations that may generalize across modalities. In this study, we investigated whether latent embeddings derived from an MRI foundation model could be applied to both MRI and positron emission tomography (PET) data for IDH mutation classification.
A total of 487 subjects from the publicly available BraTS dataset were used to train and evaluate MRI-based classification models after age- and sex-matching (102 IDH-mutant and 102 IDH–wild-type cases). External validation was performed in an independent clinical cohort of 10 glioma patients who underwent hybrid dynamic ¹⁸F-fluoropivalate (FPIA) PET/MRI. Latent feature representations were extracted using BrainIAC, a pretrained MRI foundation model. Principal component analysis was applied to reduce feature dimensionality before classification using XGBoost. Two experimental settings were investigated: (1) MRI-only models trained on BraTS and externally validated on the clinical PET/MRI cohort, and (2) multimodal PET–MRI models integrating static ¹⁸F-FPIA PET embeddings with MRI embeddings. A conventional radiomics pipeline was implemented for comparison.
In the MRI-only analysis, the combination of T1, FLAIR, and arterial spin labeling (ASL) achieved the strongest external validation performance on the mpFPIA dataset, reaching an accuracy of 0.80 and an AUC of 0.90. When PET embeddings were incorporated, embedding-based models consistently outperformed radiomics across modalities. PET-derived embeddings alone achieved an accuracy of 0.75 and an AUC of 0.86, while multimodal PET–MRI embeddings reached an AUC of 0.78 and an accuracy of 0.68. Across all configurations, embedding-based models demonstrated higher accuracy, more balanced precision–recall performance, and substantially lower feature dimensionality compared with radiomics-based approaches. Saliency analysis further confirmed that the learned representations focused predominantly on tumor regions.
These results provide the first evidence that MRI foundation model embeddings can generalize to PET imaging, enabling unified multimodal pipelines for molecular glioma characterization. Cross-modal representation learning may therefore represent a promising strategy for developing robust imaging biomarkers for non-invasive tumor genotyping.Speaker: Marianna Inglese (Department of Biomedicine and Prevention - University of Rome Tor Vergata) -
51
Spatiotemporal FPIA PET–MRI Phenotyping Reveals Oligoclonal Metabolic Niches and Vascular Remodeling in Brain Metastases
Intracranial metastatic disease (IMD) remains a major cause of cancer-related morbidity and mortality, yet the metabolic heterogeneity of brain metastases is poorly characterized in vivo. Conventional imaging typically treats lesions as metabolically homogeneous entities, potentially obscuring biologically meaningful tumor subregions. Here, we present a multimodal imaging framework integrating dynamic 18F-fluoropivalate (FPIA) positron emission tomography (PET) with multiparametric magnetic resonance imaging (mpMRI) to identify metabolic oligoclones within brain metastases and investigate their biological and clinical relevance.
Twenty-one patients (22 imaging sessions) with brain metastases from lung, breast, melanoma, and colorectal primaries underwent integrated dynamic FPIA PET–MRI either at baseline (treatment-naïve, n = 12) or 4–8 weeks after stereotactic radiosurgery (SRS, n = 10). Voxel-wise PET time–activity curves (TACs) were analyzed using unsupervised time-series k-means clustering, allowing identification of distinct metabolic subpopulations based on tracer kinetic behavior rather than static uptake metrics. PET-derived clusters were spatially mapped to tumor volumes and characterized using diffusion (ADC), perfusion (CBF, CBV), and permeability parameters (Ktrans, vp, ve) derived from dynamic contrast enhanced (DCE) and dynamic susceptibility constrast (DSC) MRI, together with PET kinetic parameters (K1, k2, k3, Ki).
In treatment-naïve lesions, clustering revealed three reproducible metabolic phenotypes: an intermediate kinetic cluster, a fast uptake/clearance cluster, and a slow-trapping cluster characterized by sustained tracer retention and positive Patlak Ki values. These oligoclones were spatially intermingled throughout the tumor volume, demonstrating microscale metabolic heterogeneity. Cluster prevalence was associated with overall survival: a higher proportion of the intermediate kinetic phenotype correlated with longer survival, whereas enrichment of the slow-trapping phenotype was associated with poorer outcomes.
Integration with multiparametric MRI revealed that these metabolic clusters correspond to distinct vascular phenotypes. Regions corresponding to the favorable phenotype exhibited higher tracer delivery, vascular permeability, and plasma volume together with shorter capillary transit times, whereas the unfavorable phenotype showed reduced delivery and permeability with prolonged transit times. Across clusters, capillary transit time demonstrated a strong inverse correlation with vascular permeability, indicating that inefficient microvascular flow is associated with reduced permeability exchange.
A logistic regression model combining cluster prevalence and vascular imaging parameters predicted poor survival with an AUC of 0.86 (leave-one-out cross-validation; permutation p = 0.038), highlighting the prognostic relevance of the identified phenotypes.
These findings demonstrate that dynamic FPIA PET combined with multiparametric MRI enables noninvasive characterization of metabolic–vascular niches within brain metastases. This approach provides a framework for imaging-based phenotyping of tumor heterogeneity and may support biologically informed treatment strategies and adaptive radiotherapy in intracranial metastatic disease.Speaker: Marianna Inglese (Department of Biomedicine and Prevention - University of Rome Tor Vergata) -
52
Quantifying Baseline Relative Permeability between PET and DCE-MRI using a DCE Flow Phantom with Total-Body PET
Introduction:
Positron emission tomography (PET) and dynamic contrast-enhanced MRI (DCE-MRI) probe complementary transport processes; $^{18}$F-FDG can undergo carrier-mediated uptake, while gadolinium chelates distribute via passive diffusion. Imaging both modalities in a shared permeability phantom isolates passive transport, allowing the establishment of a baseline permeability ratio (P$_{FDG}$/P$_{Gd}$) free of transporter effects. Measuring deviations from this passive baseline in vivo may allow transporter activity to be mapped in regions of altered vascular integrity that would otherwise confound PET analysis.Methods:
A DCE-MRI permeability phantom (Sarwar MS et al. Magn Reson Med. 2025) underwent sequential imaging with dynamic $^{18}$F-FDG PET and gadobutrol-enhanced MRI using the Siemens Biograph Vision Quadra PET and 3T Magnetom Skyra scanners.PET acquisition was gated at 5s intervals during the early dynamic phase to enable high temporal resolution measurement of tracer arrival. This allowed for estimation of flow (F) through an adiabatic approximation (St Lawrence KS, Lee TY. J Cereb Blood Flow Metab. 1998). During the adiabatic window, a sigmoid-gated plateau function was fitted to the flow curve, with plateau boundaries defined at 95% and 5% of the on and off sigmoid functions respectively. This allowed for quantification of both flow (F) and channel transit time (T$_{c}$), from which channel volume (V$_{chan}$) could be estimated using V$_{chan}$ = F*T$_{c}$.
K$_{1}$ (PET) and K$_{Trans}$ (DCE-MRI) were derived using kinetic modelling. Under the Renkin–Crone model, these parameters were combined with the measured flow to estimate the permeability–surface area product (PS). Flow was assumed to remain constant across modalities owing to the fixed experimental pump speed. A ratio of PS values was then taken to estimate the permeability ratio, P$_{FDG}$/P$_{Gd}$. A delay parameter was included to account for differences between radiotracer arrival at the measured arterial input; this also reflected differences in dispersion behaviour evident between modalities, which are thought to be due to differences in tracer viscosity. Both channel (V$_{chan}$) and pore (V$_{pore}$) volume parameters were compared against modalities and material characteristics (V$_{chan} = 0.11$; V$_{pore} = 0.75$) to validate model performance.
Results:
Volume parameter Vchan was overestimated by the flow analysis, which is consistent with the uncertain bounds of the adiabatic window. A two tissue irreversible compartment model best described the $^{18}$F-FDG kinetics, consistent with slow tracer accumulation around the phantom material over the scan timescale. Estimates of V$_{chan}$ and V$_{pore}$ were consistent across modalities and showed agreement with the phantom’s material specifications, supporting the validity of the fitted transport parameters and the derived permeability measures. The permeability ratio satisfied P$_{FDG}$/P$_{Gd}>1$, as expected given the relative molecular sizes of FDG and gadobutrol.Conclusion:
This study establishes a multimodal, multiparametric framework for quantifying PET tracer transport using a baseline permeability ratio under passive conditions. Agreement in volume parameters between modalities supports the validity of these measurements despite differences in tracer kinetics. Additional scans are planned across phantoms with varying channel and porosity dimensions to further characterise this permeability ratio and explore the concurrent dynamics between modalities. Future work will also refine the definition of the adiabatic window boundaries.Speaker: Robbie Haynes (School of Physics & Astronomy, University of Edinburgh, United Kingdom)
-
48
-
Gala Dinner
-
-
-
Session 13: TBPET-Overview & DetectorsConvener: Marta Freire
-
53
Overview of recent developments in Total Body PET
This talk will discuss key technical factors, including detector design, data processing challenges, and system cost, which is largely driven by scintillator materials and electronics. It also compares different system lengths, showing that longer scanners yield higher sensitivity but at increased cost. Novel emerging systems from GE and United imaging will be described in more detail. Finally, the talk explores strategies to reduce cost and improve performance, including alternative materials like BGO or plastic scintillators, sparse detector configurations, and advanced reconstruction methods such as deep learning.
Speaker: Stefaan Vandenberghe -
54
PET detector based on BGO pseudo-slabs: a comparison study
Recent developments in PET detectors have explored alternatives to conventional pixelated scintillator arrays in order to improve spatial resolution and DOI capabilities while maintaining good timing performance and reducing system cost. Among the different approaches proposed, semi-monolithic and pseudo-slab scintillation crystals aim to combine the advantages of pixelated and monolithic detectors by introducing controlled light sharing while preserving some degree of crystal segmentation. Moreover, BGO scintillators have regained popularity due to its lower cost, higher stopping power, higher photo-fraction and the lack of intrinsic radiation.
The main goal of this work is to compare the semi-monolithic and pseudo-slab designs using both LYSO and BGO scintillation crystals. In particular, four different configurations were tested: semi-monolithic blocks of 1×8 slabs (BGO and LYSO), a single LYSO pseudo slab, and an array of 1×8 BGO pseudo-slabs; all provided by Epic Crystals. The detectors were coupled to an 8×8 SiPM array (Hamamatsu S14) and readout using the TOFPET2 ASIC. A NN approach based on a MLP was implemented to estimate the interaction position along the monolithic and DOI direction. The models were trained with experimental data. The spatial resolution was evaluated from the error distributions of the predicted positions using the FWHM, MAE and Bias metrics.
The results show comparable DOI performance between pseudo-slab and semi-monolithic configurations, indicating that pseudo-slab detectors represent a viable alternative for PET systems requiring DOI capability. Ongoing work focuses on the analysis of the timing capabilities including the development of alternative electronic readout schemes to exploit the Cherenkov radiation of BGO crystals.Speaker: Neus Cucarella Melo -
55
Selection of plastic scintillators for the total-body J-PET scanner
Total-body Jagiellonian positron emission tomography (TB-J-PET) is based on long plastic scintillators [1] which decrease cost of the scanner [2]. Total-body PET scanners enable positronium lifetime imaging of organs in human body [3] and tissue samples [4], measurements of polarization of photons [5], CPT symmetry [6], and beam therapy monitoring [7]. Development of TB-J-PET requires application of transparent plastic scintillators with low light attenuation [8] to build long modules with silicon photomultipliers attached at both ends of the scintillators. Modular TB-J-PET construction requires quality control of plastic scintillators and verifying its optical properties [9]. Purpose of this research is to verify time resolution of plastic scintillators for the TB-J-PET modules construction.
Six types of polyvinyltoluene-based plastic scintillators with mission spectra covering the maximum quantum efficiency of light detection of silicon photomultipliers, were measured. The plastic scintillators had dimensions of 6 mm × 30 mm × 500 mm, polished surfaces: faces as-cast and edges diamond-milled, and were manufactured by Eljen Technology. The time resolution was measured at three points along the scintillator using a setup consisting of silicon photomultipliers (Hamamatsu, S13361-6674), oscilloscope, power supply, black box, and collimated Na-22 source.
The best time resolution were achieved by the EJ-204, EJ-200, EJ-208 plastic scintillators combining short signal decay time, high light output, high transparency, and the best match of the emission spectrum to the maximum quantum efficiency of the photomultipliers. The best plastic scintillator type for the next generation total-body J-PET scanner is EJ-200. The EJ-200 plastic scintillator combines the best time resolution with uniform time resolution along the scintillator strip.
[1] S. Vandenberghe, P. Moskal, J. Karp, EJNMMI Physics 7 (2020) 35
[2] P. Moskal et al., Physics in Medicine and Biology 66 (2021) 175015
[3] P. Moskal et al., Science Advances 10(37) (2024) eadp2840
[4] M. Das et al., Scientific Reports (2026) in press
[5] P. Moskal et al., Science Advances 11(18) (2025) eads3046
[6] N. Chug et al., Physical Review D 113 (2025) 032003
[7] K. Parodi, T. Yamaya, P. Moskal, Journal of Medical Physics 33 (2023) 22
[8] Ł. Kapłon, et al., Nuclear Instruments and Methods in Physics Research A 1051 (2023) 168186
[9] Ł. Kapłon et al., Journal of Instrumentation 20 (2025) P09019Speaker: Dr Lukasz Kaplon (Jagiellonian University) -
56
Sponsor Keynote: United Imaging: Harnessing Ultra-Fast TOF and Total-Body with AI: The Intelligent PET/CT Pipeline
Gold sponsor keynote talk
Speaker: Ahmadreza Rezai
-
53
-
Coffee Break
-
Session 14: TBPET-SimulationsConvener: Samuel Espana
-
57
CRYSP: towards an accessible TBPET
Nowadays, the bulk of the cost in TBPET systems comes from the scintillator, followed by SiPMs and electronics. Commercial TBPET scanners typically use LYSO, which is very expensive and relies on rare-earth elements whose production and manufacturing are geographically concentrated and supply-limited. Thus, the first priority of any putative low-cost TBPET is to find significantly cheaper alternatives to LYSO.
In this work, we propose CRYSP, a novel TBPET scanner based on large monolithic crystals of pure cesium iodide (CsI) operating at cryogenic temperatures ($\sim$100 K). Cooling pure CsI from room temperature increases its light yield by a factor of 20, reaching ~105 photons/keV, though its decay time also increases from 15 ns to 800 ns. The resulting energy resolution, below 7% at 511 keV, combined with the monolithic form factor and its inherent depth-of-interaction capability, allows millimeter-scale spatial resolution in all three dimensions, minimizing parallax error. These features more than compensate for the lower sensitivity and lack of time-of-flight, yielding performance comparable to state-of-the-art TBPET scanners at significantly reduced cost.
The slow scintillation signal makes conventional commercial PET readout solutions, such as PETsys, unsuitable for cryogenic CsI, as the integration window is too short to capture the full scintillation pulse. To overcome this, an alternative front-end based on the CAEN CITIROC ASIC was adopted. Unlike PETsys, CITIROC provides peak-hold and pulse-height analysis measurements well suited for slow scintillators, allowing reliable extraction of deposited energy. A dedicated study evaluated its performance across various sources and crystals, comparing it with the available commercial solutions, PETsys and a waveform digitizer. This demonstration of CsI’s intrinsic performance motivates the use of monolithic geometries coupled to SiPM arrays. Unlike pixelated arrays, a monolithic crystal preserves the continuous light distribution, enabling event localization in all three dimensions. A dedicated deep-learning algorithm was developed to further improve position reconstruction. Trained in PyTorch on simulated Geant4 events, the algorithm is able to identify in Monte Carlo data up to two interaction vertices per event, achieving millimeter-scale resolution in all three dimensions, including depth-of-interaction. Moreover, the algorithm will be also validated against experimental data.
As a first experimental milestone, the group is currently assembling a small-animal PET prototype composed of 18 monolithic CsI(Tl) crystals arranged in three rings, operating at room temperature. This proof-of-concept system will validate the full acquisition and reconstruction pipeline before proceeding to a fully cryogenic implementation.
Speaker: Laura Navarro-Cozcolluela (DIPC) -
58
Definitive Design for Developing a Cost-Effective Total-Body J-PET from Plastic Scintillators
Abstract:
The Jagiellonian Positron Emission Tomograph (J-PET) collaboration is developing a Total-Body PET system based on plastic scintillators, with silicon photomultipliers (SiPM) located at the strip ends, and gamma detection is achieved through Compton scattering [1]. Results presented at the PSMR 2024 conference included time-difference measurements for plastic scintillators of various lengths in order to determine an optimal detector configuration for a large axial-field-of-view (AFOV) system. Those results motivated the initial concept of constructing the Total-Body J-PET scanner using seven detector rings composed of 33 cm long scintillator modules. The main aim of the J-PET collaboration is to build a cost-effective total-body PET system that can be used for different types of studies, including two-gamma imaging [2], positronium imaging [3], and photon polarization studies [4].
In this study, we report on detector characterization studies leading to the definitive design of the Total-Body J-PET (TB-J-PET) scanner, together with a simulation-based performance evaluation for two-gamma and three-gamma imaging. Experimental measurements were performed using plastic scintillator strips read out by two types of silicon photomultipliers produced by Hamamatsu and Onsemi, together with a collimated beam of 511 keV photons from a Na-22 isotope source. Based on the experimental results, Hamamatsu SiPMs were selected due to their superior performance. With additional experiments, we obtained that by optimizing the operating voltage of the detectors, scintillators with lengths of 33 cm and 60 cm can achieve comparable timing performance. These findings enabled a redesign of the detector geometry and reduced the construction cost.
The definitive design of the TB-J-PET configuration consists of five detection rings with a total AFOV of 254 cm. Two short rings are composed of 33 cm long modules and are intended for the upper-body region, while three long rings are built from 60 cm modules, allowing a reduction in the total number of detector modules while maintaining detector performance. Each ring consists of 24 detector modules, and each module comprises three detection layers.
Simulations were conducted using GATE to evaluate the performance of the TB-J-PET scanner. The studies included both 2-gamma events and 3-gamma events originating from ortho-positronium decays. Dedicated event selection criteria and reconstruction algorithms were implemented within the J-PET Framework to process the simulated data and reconstruct interaction positions. Key performance parameters, such as sensitivity, scatter fraction, and spatial resolution, were evaluated for the TB-J-PET scanner. The simulation results indicate the possibility of the TB-J-PET to perform high-sensitivity 2-gamma and 3-gamma imaging studies.Acknowledgements:
We acknowledge support from NCN (grants 2021/42/A/ST2/00423, 2021/43/B/ST2/02150,
2022/47/I/NZ7/03112), MNiSW (IAL/SP/596235/2023 and SPUB/SP/627733/2025), SciMat and qLife Priority Research Areas, ERC Advanced Grant POSITRONIUM no. 101199807, and PLGrid infrastructure (PLG/2024/017688, PLG/2025/018762).Reference:
[1]. P. Moskal et al., PET Clinics, vol. 15, no. 4, pp. 439-452, 2020.
[2]. M. Das et al., Bio-Algorithms and Med-Systems, vol. 20, no. 1, pp. 101-110, 2024.
[3]. P. Moskal et al., Sci Adv, vol. 10, no. 37, pp. eadp2840, 2024.
[4]. P. Moskal et al., Sci Adv, vol. 11, no. 18, pp. eads3046, 2025.Speakers: Keyvan Tayefi Ardebili (Jagiellonian University), Prof. Pawel Moskal (Jagiellonian University) -
59
Toward Total-Body Modular Flat-Panel TOF-PET Enabled by Sub-100 ps Timing
Total-body PET (TB-PET) offers transformative gains in sensitivity, dynamic imaging capability, and quantitative performance, but current implementations rely on extended axial full-ring geometries associated with high scintillator cost and system complexity. Within the PetVision project, we investigate an alternative paradigm in which extreme timing performance enables simplified, modular detector geometries scalable toward the total-body regime.
The PetVision concept is based on opposing planar detector modules composed of fast pixelated scintillators, low-crosstalk NUV-sensitive SiPM arrays, and high-bandwidth ASIC readout with on-chip digitization. The architecture is intrinsically modular: detector panels can be extended axially and tiled to progressively approach TB-PET configurations, while maintaining flexibility in system geometry and deployment.
The central hypothesis is that coincidence time resolution (CTR) below 100 ps can compensate for limited angular coverage and enable high-quality image reconstruction in sparse detector layouts. Monte Carlo simulations and reconstruction studies demonstrate that as timing performance improves, limited-angle artefacts are strongly suppressed, and planar configurations approach the image quality of conventional clinical systems despite significantly reduced detector material.
Crucially, recent single-pixel timing performance confirms the feasibility of this approach, demonstrating that the required timing precision can be achieved at the detector level. Furthermore, we report experimental results obtained with a predecessor ASIC to the FastIC+ chip, which will be used in the PetVision demonstrator. These measurements validate the capability of the readout chain to preserve ultrafast timing information and support the targeted sub-100 ps system performance.
In addition to timing, system scalability is addressed through high-density integration concepts, including 2.5D interconnect architectures that minimize parasitic effects and enable compact module design. Combined with ongoing developments in low-crosstalk SiPMs and optimized scintillators, this provides a realistic pathway from laboratory demonstrators to large-area detector panels.
This work outlines a new route toward TB-PET based on ultrafast timing, modular manufacturability, and reduced material usage. Rather than scaling conventional ring geometries, the PetVision approach enables total-body coverage through detector tiling and timing-driven reconstruction. If validated at system level, this concept could significantly lower the barrier to TB-PET deployment and broaden access to high-performance molecular imaging.
Speaker: Rok Pestotnik (Jozef Stefan Institute (SI)) -
60
Towards Autonomous Position Calibration of PET Scanners using Physics-Based Neural Networks
The recent advent of Total-Body PET (TB-PET) represents a transformative leap in molecular imaging, offering an effective sensitivity gain of over 40-fold and enabling full-body dynamic tracking [1, 2]. To maintain uniform spatial resolution across such extended fields of view, precise Depth-of-Interaction (DOI) capability is crucial to mitigate severe parallax errors. Advanced light-sharing detector designs ranging from multi-layer finely segmented arrays to fully monolithic scintillation crystals provide excellent intrinsic 3D positioning capabilities to meet this demand [3]. However, the translation of these high-performance architectures into large-scale clinical TB-PET systems is severely hindered by a critical bottleneck: the prohibitive complexity of their positioning calibration. Traditional 3D calibration relies on meticulous, labor-intensive benchtop setups using mechanically scanned collimated gamma-ray sources [4].
While recent in-system "virtual collimation" methods have shown promise in scaling this process for fully assembled scanners [5, 6], they fundamentally rely on pre-existing, accurate position estimates in a partner 'collimating' detector to define a precise Line-of-Response (LOR). In a newly assembled system where all detectors are uncalibrated, this creates a circular dependency. Relying on uncalibrated analytical methods for initial estimates propagates significant spatial uncertainties, preventing true de novo calibration [7].
To overcome this fundamental barrier, we present a novel, autonomous in-system calibration methodology driven by Physics-Informed Neural Networks (PINNs) [8]. Rather than relying on physical collimators or imprecise analytical pre-positioning, our approach leverages the fundamental physics of positron annihilation as an implicit, powerful training constraint . By placing a simple, uncollimated source at various locations within the scanner’s field of view, the neural network learns to directly map raw electronic detector signals to 3D interaction coordinates based purely on the physical laws dictating that the coincident gamma-rays must geometrically intersect the known source position.
In this work, we present the evolution of this methodology, beginning with an ideal proof-of-concept case. Using simplified Monte Carlo simulations restricted to single-point photoelectric absorptions, we demonstrate the baseline capability of the physics-constrained network to achieve highly accurate 3D positioning (sub mm) from scratch, without any prior analytical pre-positioning or ground-truth label generation.
Following this ideal case, we detail the progressive introduction of realistic detector phenomena and the necessary adaptations to our AI framework. We transition to high-fidelity Geant4 simulations of light-sharing scintillators, introducing multi-interaction events such as Compton scattering.
By detailing this progression from idealized simulations to handling the noisy, multi-variate reality of physical detectors, we demonstrate a clear pathway toward experimental validation. Ultimately, this physics-based deep learning framework promises to replace weeks of mechanical calibration with rapid, autonomous de novo calibration and routine fine-tuning, unlocking the cost-effective scaling required for the future of TB-PET and advanced multimodality integrated systems.
References:
[1] Vandenberghe., et al. EJNMMI Phys 7 (1) (2020): 35.
[2] Spencer, Benjamin A., et al. J Nucl Med 62 (6) (2021): 861-870.
[3] Gonzalez-Montoro, Andrea, et al. IEEE TRPMS 5 (3) (2021).
[4] Berg, Eric, and Simon R. Cherry. Semin Nucl Med 48 (4) (2018).
[5] Bruyndonckx, Peter, et al. NIM-A 571 (1-2) (2007).
[6] Gonzalez-Montoro, Andrea, et al. IEEE TRPMS 5 (6) (2020).
[7] Kuhl, Yannick, et al. Med Phys 52 (1) (2025): 232-245.
[8] Cai, Shengze, et al. Acta Mech Sin 37 (12) (2021): 1727-1738.
Speaker: PABLO GALVE LAHOZ (Institute for Physical and Information Technologies “Leonardo Torres Quevedo”, ITEFI, Spanish National Research Council (CSIC), Madrid, Spain) -
61
Investigation of the Sensitivity of a Brain PET Insert in a Total-Body J-PET System: A Simulation Study
Positron emission tomography (PET) is a powerful molecular imaging tool that detects functional and biochemical changes in tissues before structural abnormalities become visible,making it invaluable for early diagnosis, disease characterization, and treatment planning and monitoring [1]. Sensitivity is a critical determinant of PET image quality and is significantly enhanced in systems with a wide axial field of view (AFOV).
This study aims to evaluate and compare the sensitivity of a total-body J-PET system and a plastic-based brain PET scanner, as well as to investigate the combined sensitivity achieved when the brain scanner is integrated into the whole-body system.
Monte Carlo simulations were performed using the GATE platform to model the Jagiellonian Positron Emission Tomograph (J-PET), a next generation total-body PET prototype with an AFOV exceeding 250 cm, constructed from cost-effective plastic scintillators rather than conventional crystal-based detectors [2, 3].
The simulated sensitivity results for both standalone and integrated configurations will be presented at the upcoming conference, providing insight into the performance potential of plastic scintillator based PET systems for broad clinical and research applicationsSpeakers: Prof. Pawel Moskal (Jagiellonian University), sharareh jalali
-
57
-
Lunch Break
-
Session 15: TBPET-MethodsConvener: Francis Loignon-Houle
-
62
A multi-axes gantry system for total-body J-PET/CT scanner
Introduction
Conventional PET/CT systems acquire images sequentially along a single axis using a moving patient table, typically within one or two adjacent gantry enclosures. Although whole-body imaging is feasible in this configuration, it requires multiple discrete bed positions and is therefore more susceptible to motion-related artefacts than total-body approaches. In addition, total-body imaging offers the potential to image the same volume with reduced radiopharmaceutical activity compared with standard multi-bed whole-body protocols [1].Methods
The total-body Jagiellonian-PET (J-PET) project [2,3] adopts a two-axis architecture in which PET and CT operate around the same stationary patient table from different motion axes. This concept is intended to reduce involuntary patient-motion artefacts while maintaining a compact system layout. However, the implementation of crossed-axis motion for heavy imaging modules introduces demanding mechanical requirements, including the transport of components exceeding 1000 kg with sub-millimetre positioning precision on intersecting axes within a shared footprint. To address these constraints, we developed a cross-stage sliding gantry concept based on discrete rail placement and lead-screw-driven stages, avoiding conventional stacked linear-motion arrangements [4]. The motion platform is built on a custom aluminium-alloy base frame and incorporates hardened steel ball screws, two servo motors, and two servo drivers for two-axis gantry motion. In parallel, a dedicated vertical-motion system was developed for the patient table and integrated at both ends of the support. The table consists of a lightweight sandwich structure composed of 3 mm carbon fibre, 2.4 cm PMI foam, and 3 mm carbon fibre, thereby reducing both weight and photon interaction within the patient support. The resulting system enables examinations over a scanning length of up to 2.5 m, sufficient for full-body PET studies.Results
The 3D design, fabrication, and installation of the two-axis gantry system have been completed successfully. The next phase will include CT image acquisition using the installed platform, followed by integration of the total-body J-PET detector for combined anatomical and metabolic imaging.Conclusion
The two-axis sliding gantry system has been installed at the Theranostics Center in Krakow. We present the implemented mechanical architecture and operational workflow for compact total-body J-PET/CT imaging around a stationary patient support. The current configuration demonstrates the feasibility of dual-axis multimodal imaging with reduced dependence on patient-table translation and provides a flexible platform for further development of total-body J-PET/CT.Acknowledgements
We acknowledge support from NCN (grants 2021/42/A/ST2/00423, 2021/43/B/ST2/02150, 2022/47/I/NZ7/03112, 2023/50/E/ST2/00574), MNiSW (IAL/SP/596235/2023 and SPUB/SP/627733/2025), SciMat and qLife Priority Research Areas, ERC Advanced Grant POSITRONIUM no. 101199807.[1] Vandenberghe, S., et al., (2020). State of the art in total body PET. EJNMMI Physics, 7, 35.
[2] Moskal, P. et al. (2021). Simulating NEMA characteristics of the modular total-body J-PET scanner—an economic total-body PET from plastic scintillators. Physics in Medicine & Biology, 66(17), 175015.
[3] Moskal, P., & Stępień, E. Ł. (2020). Prospects and Clinical Perspectives of Total-Body PET Imaging Using Plastic Scintillators. PET Clinics, 15(4), 439–452.
[4] Kaplanoglu, T., & Moskal, P. (2023). Across-staged gantry for total-body PET and CT imaging. Bio-Algorithms and Med-Systems, 19(1), 109–113.Speaker: Tevfik Kaplanoglu -
63
Studies of attenuation and detector geometry effects in two- and three-gamma positronium decays towards 3γ/2γ imaging using the Total-Body J-PET tomograph.
Conventional Positron Emission Tomography (PET) relies on the detection of two back-to-back 511 keV photons originating from electron-positron annihilations. However, about 40 percent of annihilations occur via positronium formation, including ortho-positronium (o-Ps) decays into three photons. The features of positronium, including its lifetime and the ratio of 3γ to 2γ annihilations (3γ/2γ), strongly depend on the architecture of the surrounding tissue. This has led to the development of positronium imaging – an emerging PET extension that provides additional information about the imaged tissue and may improve the diagnostic process.
In contrast to the conventional PET devices, the Jagiellonian Positron Emission Tomography scanner (J-PET), based on plastic scintillators, is capable of multi-photon detection. This feature enables the registration of 3γ events, making it able to assess the properties of positronium, such as its lifetime and the 3γ/2γ decay ratio, and utilize these properties in medical imaging. The 3γ/2γ method, in contrast to positronium lifetime imaging, does not require the registration of the de-excitation photon, meaning that all commonly used PET radioisotopes are suitable for this approach. The method can be applied with standard radiopharmaceuticals, such as 18F-FDG, and with emerging isotopes like 44Sc, which we plan to utilize in J-PET.
Accurate tomographic image reconstruction requires corrections for attenuation, scatter, and random coincidences. Such methods are currently implemented in the J-PET detector for conventional 2γ imaging. However, they are not suitable for the 3γ approach. This limitation creates the need to develop correction methods dedicated to this technique.
In this work, we investigate absorption effects for both 2γ and 3γ annhilation processes using Monte Carlo (MC) simulations. Studies were performed with a custom toy MC implemented in ROOT and with the GATE simulation toolkit for a range of phantom models, including simplified geometries and the anatomically realistic XCAT phantom. The results are presented in the form of absorption maps for para- and ortho-positronium decays that can serve as a foundation for attenuation correction.
Moreover, preliminary simulations including the Total-Body J-PET detector geometry were performed to assess the role of detector effects, including detector acceptance and energy thresholds relevant for signal registration, in the feasibility of 3γ/2γ positronium imaging. The results show significantly lower detection efficiency for 3γ events compared to 2γ, due to higher attenuation of low-energy photons and limited detector acceptance. The efficiency is also strongly dependent on the energy threshold, with even small increases causing a substantial reduction in the number of detected 3γ events, while 2γ detection remains largely unaffected.Speaker: Kamila Kasperska (Jagiellonian University, Kraków, Poland) -
64
From image to video in Total-Body PET: Maximum-Likelihood Motion and Activity (MLMA) reconstruction
The advent of long axial field-of-view (LAFOV) PET has enabled simultaneous imaging of all organs with substantially higher sensitivity. However, motion remains an issue that degrades image quality, as the sensitivity of LAFOV-PET is insufficient to mitigate fast irregular motion at the subsecond timescale. We therefore propose to correct for motion in LAFOV-PET at high frequency (2 Hz) by directly processing list-mode data with the Maximum-Likelihood Motion and Activity (MLMA) reconstruction method, optimized for the large projection space of LAFOV-PET through data reduction. We demonstrate the method on the NEMA body phantom scanned at the IMAS Total-Body PET scanner (I3M Valencia, Oncovision). The results indicate that MLMA can be efficiently applied to LAFOV-PET, realizing the full potential of these high-sensitivity scanners.
Speaker: Rodrigo José Santo (UMC Utrecht) -
65
DSAC: A Novel CT-less Attenuation Correction Approach for Cost-Effective Plastic Scintillator based Total-body PET
Background: The development of Total-Body PET (TB-PET) based on plastic scintillators, such as J-PET [1,2], offers a longer axial field-of-view and cost-effective framework for whole-body imaging. This technology also enables advanced applications like positronium imaging of the human brain [3]. The necessity of an external CT scanner for attenuation correction (AC) limits the benefits of TB-PET by adding the cost, mechanical complexity & radiation exposure. In high-sensitivity TB-PET systems, the reduction in required tracer dose means the CT component now accounts for the majority of the patient's radiation exposure [4, 5]. We propose Detector-Scattered Attenuation Correction (DSAC) [6], which is a hardware-less method for CT-less AC. This framework converts inter-detector scattering, traditionally treated as noise, into a signal. The core idea is that as detector-scattered photons traverse the patient's body, they carry tissue information that can be used to construct an attenuation map.
Methods: We developed custom Monte Carlo simulations for a plastic scintillator-based total-body J-PET geometry to track photon transport through biological tissue models. Our analysis focused on triple-hit coincidence events, where an annihilation photon interacts with one detector module, undergoes Compton scattering, and is subsequently detected in a second module. At the same time, the second annihilation photon from the pair travels in the opposite direction and hits a third module. Due to the low atomic number of plastic scintillators, these scattering events occur frequently [1, 2]. For each event, the system recorded spatial coordinates and Time-of-Flight (TOF) data, which were then processed by a reconstruction algorithm. This algorithm utilized the spatial information to estimate the line-integral of the linear attenuation coefficient, enabling the mapping of material density without external anatomical data.
Results: The simulation results demonstrate the technical feasibility of the proposed approach. The algorithm successfully reconstructed tissue density maps independent of external CT datasets. A strong linear correlation was observed between the calculated density values and the ground-truth densities of the simulated test objects. These findings validate that the high probability of Compton scattering within plastic scintillators provides a viable signal for attenuation estimation. Furthermore, the extended axial field-of-view (AFOV) characteristic of total-body scanners significantly enhances the geometric sensitivity and capture rate of these inter-detector scattering events.
Conclusion: DSAC offers a viable, hardware-less method for plastic-scintillator based TB-PET systems. By eliminating the necessity for integrated CT hardware, this method reduces overall system complexity, lowers manufacturing costs, and minimizes patient radiation burden. Future research will focus on the optimization of image quality and the implementation of advanced denoising techniques to manage the statistical variance inherent in scatter-based data due to limited statistics.
References:
1. Moskal P, et al. Nucl Instrum Methods Phys Res A. 2014;764:317-321. doi:10.1016/j.nima.2014.07.010
2. Moskal P, et al. PET Clin. 2020;15:439-452. doi:10.1016/j.cpet.2020.06.012
3. Moskal P, et al. Sci Adv. 2024;10:adp2840. doi:10.1126/sciadv.adp2840
4. Badawi RD, et al. J Nucl Med. 2019;60:299-303. doi:10.2967/jnumed.118.221226
5. Cherry SR, et al. J Nucl Med. 2018;59:3-12. doi:10.2967/jnumed.116.184028
6. Tiwari S, et al. Bio-Algorithms Med-Syst. 2025;21:13-20. doi:10.1515/bams-2024-0010Acknowledgement: We acknowledge support from the National Science Centre (NCN), grant nos. 2021/42/A/ST2/00423, 2021/43/B/ST2/02150, 2022/47/I/NZ7/03112, and 2023/50/E/ST2/00574; the Polish Ministry of Science and Higher Education (MNiSW), grants no. IAL/SP/596235/2023 and SPUB/SP/627733/2025; the SciMat and qLife Priority Research Areas; and the ERC Advanced Grant POSITRONIUM (no. 101199807).
Speaker: Satyam Tiwari -
66
High throughput imaging with Walk-Through-PET
The Walk-Through PET (WT-PET) is an innovative approach to Positron Emission Tomography imaging that rethinks both system geometry and patient workflow. Unlike conventional cylindrical PET scanners that require patients to lie still on a moving bed, the WT-PET uses two vertically aligned flat detector panels positioned on either side of a standing patient. This design introduces a more open and accessible scanning environment, allowing patients to walk into the system and position themselves without extensive assistance. The scanner dimensions—106 cm in height, a 50 cm adjustable gap, and 71 cm width—are carefully optimized based on ergonomic studies and clinical datasets to ensure both comfort and imaging performance.
A major advantage of the WT-PET system lies in its ability to significantly streamline clinical workflow. By eliminating the need for a patient bed and minimizing staff involvement in positioning, the system reduces setup time and increases overall throughput. With scan durations as short as 30 seconds and the ability to handle up to 12 patients per hour, WT-PET has the potential to dramatically improve efficiency in busy imaging centers. This is particularly valuable in regions with high demand for PET imaging, where scanner availability is often a limiting factor.
From a technical perspective, the system employs monolithic detectors capable of high spatial resolution, depth-of-interaction (DOI) measurement, and time-of-flight (TOF) capabilities. These features contribute to more uniform image resolution—below 2 mm—across the imaging field, which is an improvement over many traditional systems. Despite using approximately 50% less detector material due to its planar configuration, the WT-PET maintains competitive sensitivity (around 120 kcps/MBq), approaching that of larger cylindrical long axial field-of-view scanners. This reduction in detector area not only lowers system cost but also contributes to a smaller physical footprint, making installation more feasible in constrained clinical environments. Another important benefit is the more efficient use of radiotracers. The combination of high sensitivity and rapid acquisition enables a reduction of approximately 66% in the required tracer activity per patient, which has implications for cost savings. Further savings on system cost were investigated by simulating sparse configurations.
However, the WT-PET concept also introduces several challenges. Patient motion is a concern when scanning in an upright position, but this is mitigated through rapid acquisition times, ergonomic supports such as headrests and adjustable hand grips, and real-time feedback systems. The absence of an integrated CT scanner for attenuation correction is addressed with deep learning–based methods that estimate attenuation maps directly from PET data. Additionally, the planar geometry results in limited-angle sampling, which can affect image reconstruction. This is compensated for by leveraging the system’s large axial acceptance angle, along with DOI and TOF information, supplemented by advanced reconstruction algorithms and AI-based corrections. Following extensive simulation studies and reconstruction optimization, the WT-PET system is currently under construction, representing a promising step toward more efficient, accessible, and patient-friendly PET imaging. A 2-row bench top has been set-up and first lab measurements are ongoing.Speaker: Stefaan Vandenberghe
-
62
-
Coffee Break
-
Session 16: TBPET-SystemsConvener: Stefaan Vandenberghe
-
67
NEMA NU-2 performance tests of a prototype PET scanner using novel detectors with measured depth-of-interaction correction
Background
Depth-of-interaction (DOI) is one of the remaining challenges for high resolution PET. The resolution degradation stems from incorrect LOR endpoint estimation when photons impinge on the detector at oblique angles. This problem increases with distance from the symmetry axis, and poses a challenge for LAFOV PET, where more LOR are acquired at oblique angles. Terapet SA manufactures radiation detectors capable of localising gamma ray interactions in 3D within the detector volume. The aim of this project was to characterise the performance of a prototype PET scanner, called the Nuclyscan prototype, using Terapet’s proprietary detectors. In particular, the reduction of DOI artefacts was of interest.
Method
The NEMA NU-2 performance tests were conducted at the Department of Nuclear Medicine at Inselspital, Bern, Switzerland. Additionally, the spatial resolution test was performed in one of the clinic’s PET scanners with pixelated crystals, at the same positions relative to the edge of the FOV as in the prototype to isolate the performance difference of the detector technologies in similar geometrical configurations.
Results
The prototype achieved a FWHM of 2.56, 2.60 and 2.96 mm at 10, 100 and 200 mm radial offset, respectively. The clinical PET scanner achieved a FWHM of 3.42, 3.86 and 4.14 mm at equivalent positions. Hence, the resolution improved up to 33% in the prototype.
The prototype in its current configuration has a low detection sensitivity. This is explained by the short axial field-of-view of the prototype, the narrow energy window and restrictive coincidence-selection policy to reject scattered and random events.
The image quality assessment has not yet been completed, as the image reconstruction algorithm is undergoing further optimization before final analysis.
Conclusions
The prototype achieved the goal of demonstrating improved spatial resolution with the proprietary detectors. In particular, the spatial resolution is maintained with increasing radial offset. Future work will be aimed at improving the image reconstruction algorithms and correction methods. Construction of the full field-of-view Nuclyscan scanner using the same proprietary detectors is ongoing.Speaker: Mr August Blomgren (Department of Nuclear Medicine, Inselspital, Bern) -
68
IMAS: Long-Axial FOV PET with Simultaneous TOF and DOI Capabilities
Backgrounds and Aims
Total-body (TB) PET systems have significantly increased system sensitivity compared with conventional whole-body PET scanners. However, currently available clinical TB-PET systems are based on pixelated crystals without depth-of-interaction (DOI) capability, which may limit spatial resolution homogeneity at large radial offsets due to parallax errors. We present the initial performance evaluation and pilot clinical results of IMAS, a 71-cm axial FOV TB-PET prototype that combines simultaneous time-of-flight (TOF) and DOI capabilities using semi-monolithic LYSO detectors. The aim of this work is to describe the system design, calibration strategy, performance characterization inspired by NEMA NU 2-2018, and preliminary clinical validation.
Materials and Methods
The IMAS system consists of five 10-cm detector rings separated by 5-cm gaps, providing 71 cm axial coverage and an 82 cm bore diameter. Detector blocks are based on 3 × 25 × 20 mm³ LYSO semi-monolithic slabs arranged in 1×8 arrays and coupled to 8×8 SiPM matrices. A proprietary 64-to-16 multiplexed readout preserves 3D positioning and timing information. Data acquisition is performed with PETsys electronics, incorporating dual DAQ boards and an FPGA-based energy filter to improve count-rate capability.
Spatial positioning in the monolithic (y) and DOI (z) directions is estimated using multilayer perceptron (MLP) neural networks trained on reference super-modules and extended to all detector blocks. Timing skew correction is performed iteratively using a rotating FDG-filled bar and validated with a ²²Na point source.
Image reconstruction is performed using OSEM (5 iterations, 5 subsets), including DOI correction, TOF, attenuation correction and scatter correction. System performance was evaluated using NEMA NU 2-2018-inspired procedures, including spatial resolution, sensitivity, count-rate performance (NECR), energy resolution, coincidence time resolution (CTR), and image quality using a torso phantom. Preliminary clinical comparison was performed against a Philips Gemini TF-64 PET/CT scanner in a patient with multiple lesions.
Results
Detector-level measurements showed a detector time resolution of 196 ± 13 ps and an energy resolution of 11 ± 1%. After skew correction, system CTR improved from 2863 ps (uncorrected) to 565 ± 5 ps FWHM for a ²²Na point source. At the NECR peak (3.26 kBq/mL), CTR was 690 ps with an energy resolution of 12.8%.
The system achieved homogeneous spatial resolution across the full FOV. At the axial center, radial FWHM was 3.34 mm at 1 cm offset and 3.53 mm at 30 cm. DOI correction prevented radial mispositioning errors up to 6 mm at 30 cm off-center. The average spatial resolution was 3.3 ± 0.5 mm.
Measured sensitivity was 56.54 cps/kBq, corresponding to a peak sensitivity of 7.6% at the scanner center. The peak NECR was 79 kcps at 3.26 kBq/mL, lower than other LAFOV systems, likely due to data-transfer limitations. CTR degraded moderately with activity concentration, ranging from 550 to 725 ps over the tested range.
Preliminary clinical imaging demonstrated improved lesion delineation and higher apparent signal-to-noise ratio compared with the conventional PET/CT system. DOI capability improved lesion localization at radial distances up to 16 cm from the scanner center.Speaker: Álvaro Anreus Valero -
69
High-Resolution and Dynamic Imaging Capabilities of a 200-cm DOI-Enabled Total-Body PET System
Total-body positron emission tomography (PET) has attracted increasing attention due to its potential for significantly improved sensitivity and whole-body dynamic imaging compared with conventional PET systems. In this work, we develop a depth-of-interaction (DOI)-enabled total-body PET system with a 200 cm axial field of view (AFOV). The system employs LYSO crystal arrays with a crystal size of 2 × 2 × 20 mm³ coupled to multi-pixel photon counter (MPPC) arrays, and light-sharing windows are used to enable DOI estimation and reduce parallax errors.
The imaging performance of the system was evaluated using several phantoms, including an SZBL logo phantom, a Derenzo phantom, and a one-meter dynamic line source. The reconstructed images demonstrate that structures with dimensions down to 2 mm can be clearly resolved, confirming spatial resolution consistent with the crystal pitch. Dynamic imaging experiments further show clear propagation of 18F-FDG along the line source with high signal-to-noise ratio using 5-s frames, indicating the high sensitivity of the system. In addition, a time resolution of approximately 320 ps was achieved for both single-ring and double-ring configurations.
These results demonstrate that the proposed system provides high spatial resolution, good timing performance, and strong capability for dynamic whole-body imaging. The system shows promise for low-dose imaging and multi-organ dynamic studies in future clinical and research applications.
Speaker: Dr Qiyu Peng -
70
Evaluation of Quality Control Approaches for Monitoring DOI Correction Stability in a Long Axial FOV PET System
Background: IMAS is a long axial field-of-view (LAFOV) positron emission tomography (PET) system with time-of-flight (TOF) and depth-of-interaction (DOI) capabilities. This study aims to develop and demonstrate a practical quality control (QC) method to monitor the stability and reproducibility of DOI correction in the IMAS.
Methods: System stability was assessed using both $^{22}\mathrm{Na}$ point sources and capillary sources. $^{22}\mathrm{Na}$ sources mounted in a clinical PET/CT alignment holder were measured in three acquisitions over multiple days to evaluate reproducibility. Stability was quantified by comparing relative distances between reconstructed sources after DOI correction. Capillary sources prepared according to the NEMA NU 2 protocol were used to assess spatial resolution via FWHM in radial direction.
Results: Relative point-source positions remained largely consistent across the three acquisitions. The largest deviation of 1.94 mm occurred after holder repositioning, while differences between acquisitions without repositioning remained below 1 mm. Absolute position comparisons showed small deviations from the experimentally determined ground-truth positions, generally within 2–3 mm. All DOI-corrected reconstructed positions remained within the assumed 5 mm ground-truth uncertainty. Spatial resolution measurements confirmed previously reported FWHM behavior, with DOI correction improving radial resolution and reducing degradation at larger radial offsets.
Conclusion: The proposed QC approach enables practical monitoring of DOI correction stability in the IMAS LAFOV PET system. The results demonstrate stable DOI-corrected source positioning across repeated measurements and confirm expected spatial resolution performance. This method provides a simple and robust procedure for routine system quality control and long-term performance monitoring. Future studies will employ a dedicated holder specifically designed for this QC procedure.
Reconstructed point-source positions across three acquisitions at different radial distances from the center. Red markers (three per source) show reconstructions without DOI correction, while yellow markers show reconstructions with DOI correction. Blue markers indicate the experimentally determined ground-truth positions with an assumed uncertainty of 5 mm (dashed circle). All DOI-corrected reconstructions fall within the assumed ground-truth uncertainty.
Speakers: Carla Gonzalez Avilés, Sophia Amelie Dietrich
-
67
-