Following the series of previously conducted symposia on dedicated medical imaging instrumentation and applications, we are presently organizing the 7th issue of this event in Portorož, Slovenia. Unfortunately due to the Covid-19 pandemics, the workshop has been postponed twice. We hope to meet you in-person this year.
The MEDAMI workshops are unique in that they address a medical imaging topic from the medical, clinical, and instrumentation perspective. They foster synergistic interactions between cutting edge applications and latest technical developments. Stakeholders from academia, medical institutions and organizations, regulatory agencies, and industry have an opportunity to jointly discuss strategic directions of the medical imaging field.
Some of the key topics this year’s forum shall address are:
Alzheimer's and Parkinson's disease represent two of the most prevalent neurodegenerative diseases with dramatic socio-economic impacts for governments and healthcare systems to address. Currently, excepting the modest effects of Aducanumab in AD, there are no disease modifying therapies for either disease. However, the development of future disease modifying drugs represents a tremendously active area for the pharmaceutical industry with a large number of clinical trials under way. Imaging biomarkers are central to these clinical trials and provide important readouts of relevant disease pathology and the impact of novel therapies. This talk will review the biomarkers that are currently deployed in clinical trials and those that are in development for these devastating neurodegenerative diseases.
Molecular imaging is key in dementia, especially for early diagnosis, but also for identifying reversible forms of dementia. The objective of this presentation is to give the clinical perspective of molecular imaging for improving the early diagnosis of dementia and to explain how it may help to improve the diagnosis of reversible dementia. We will use the recent development of a new clinical predementia syndrome called the Motoric Cognitive Risk syndrome (MCR) to demonstrate how molecular imaging may help to better understand the pathophysiology of MCR. Then, we will focus on the leading cause of reversible dementia – normal pressure hydrocephalus – and show how molecular imaging may help to identify normal pressure hydrocephalus from its mimics.
PET imaging allows the production of high resolution images of radiopharmaceutical distribution in humans and animals. With the use of appropriate radiopharmaceuticals combined with tracer kinetic modeling, relevant physiological parameters such as volume of distribution (VT) and binding potential (BPND) can be quantified in vivo. The development of appropriate modeling methodology typically requires dynamic scans plus arterial input function measurement including assays of radiolabeled metabolites. Ideally, additional validation studies are also performed, including in vivo blocking studies and in vivo/ex vivo/in vitro correlations. Ultimately, to facilitate clinical utility, simplified methodology is developed, e.g., calculating standard uptake values (SUV) and their ratios (SUVR) to an appropriate (or useful) reference region. That process produces a clinically feasible approach but may suffer from increased variability or biased values which can differ in various patient populations or in different patient states. This presentation will include 1) basic PET quantification strategies for brain targets, 2) tracer validation and optimization, using as an example 11C-UCB-J, the novel SV2A agent used to assess synaptic density, 3) characterizing and optimizing a simplified quantification method using SUV and SUVR, including the inherent assumptions of these approaches, 4) examples of imaging paradigms where simplified approaches can provide misleading results, and 5) methods to improve simplified approaches without extending acquisition times to account for effects of plasma clearance of radiopharmaceutical.
The emergence of new technologies in the domain of nanophotonics, microelectronics, artificial intelligence, etc… open new perspectives for PET scanners with ultimate Time-of-Flight (TOF) performance. An ultimate goal of 10ps has been set in the frame of the 10ps TOFPET challenge: https://the10ps-challenge.org
Achieving this ambitious goal would allow improving the PET effective sensitivity by a factor of 20 as compared to the best performing TOFPET today, the Biograph Vision from Siemens, opening the way to a reduction of the radiation dose (currently 5-25 mSv for whole body PET/CT), scan time (currently > 10 minutes), and costs per patient (currently > 1000 € per scan), all by an order of magnitude.
A first generation of metascintillators has been developed with a TOF resolution of 100ps at the system level, combining the high stopping power of BGO or LYSO crystals with the ultrafast emission of cross-luminescent BaF2 or plastic scintillators.
This talk will describe a brainPET project, which will introduce this ground-breaking imaging technologies and AI image reconstruction techniques that push the limits of sensitivity and spatial resolution to better assess brain beta amyloid plaques and characterize mild cognitive impairment with the goal to delay its progression to dementia by detecting unprecedented much smaller amount of beta amyloid in the brain grey matter.
Optimization of the timing resolution in the scintillation light readout has been one of the most important challenges in the SiPM field since the beginning of their development. Several sensor parameters contribute to the timing performance achieved in the application. The latest iteration of the NUV-HD SiPM technology developed at FBK feature Photon Detection Efficiency (PDE) in excess of 60% at 410 nm, Dark Count Rate around 60 kHz/mm2 and Single-photon Time resolution (SPTR) of 90 ps FWHM for a 4x4 mm2 device with 40 um cells, when coupled to a discrete, high-frequency readout. Thanks to these parameters, it was possible to measure an excellent Coincidence Resolving Time (CRT) of 58 ps FWHM in the readout of a 2x2x3 mm3 LSO:Ce:Ca coupled to a 4x4 mm2 SiPM with 40 µm cells (98 ps FWHM with a 2x2x20 mm3 LSO:Ca:Ce). Ongoing developments include the use of metal-filled deep trench isolation, which allows reducing the optical crosstalk probability to 10% with a PDE of 60% (bare die). On the other hand, photon-starved applications, such as BGO readout with the timing resolution enhanced by the detection of Cherenkov photons, further underline the importance of improving the SPTR of the SiPMs. In the current generation of devices, this parameter is heavily affected by both the output capacitance of the sensor and by the characteristics of the front-end electronics reading it. Considering that incremental improvements between subsequent generations of SiPMs are reaching saturation, a deeper redesign of the device structure is needed. In this context, FBK is working on the development of the next-generation of SiPMs, with a strong focus on 3D integration, such as SiPMs featuring fine-pitch Through Silicon Vias and Backside-illuminated (BSI) devices. A fine segmentation of the sensitive area in separated mini-SiPMs, each one connected to a dedicated readout channel through a low-impedance interconnection, will reduce output capacitance and optimize signal integrity. BSI-SiPMs will potentially bring additional advantages, such as reaching a PDE close to 100%, reduced output capacitance, enhanced radiation hardness, single-cell connection to the readout electronics and a uniform light entrance window, suitable for the most advanced optical stacks. In the presentation, FBK roadmap towards 3D integrated SiPMs and the preliminary result obtained so far will be discussed.
The reduction of the Coincidence Time Resolution (CTR) down to a few tens of ps is one of the main challenges of basic ToF-PET research. Reaching these limits with modules that can be easily scalable to large scanners requires an optimization of the scintillator crystals, photodetectors and readout electronics employed. This work is focused on the FastIC, an 8-channel ASIC that was designed targeting ToF-PET scanners with sub-100 ps CTR, compact and scalable electronics and relatively low power consumption (∼12 mW/ch). The FastIC is meant to process fast signals and output a precise time stamp and a linear energy measurement. Moreover, it allows to readout the channels individually or to sum the outputs in groups of 4 channels. We present the performance of the FastIC ASIC when used to readout silicon photomultipliers (SiPMs) coupled to scintillator crystals. We were able obtain our lowest CTR value of (76 ± 2) ps using a 2 × 2 × 3 mm3 LSO:Ce:0.2\%Ca crystal coupled to a FBK NUV-HD-LF SiPM of 3.12 × 3.12 mm2. We discuss the performance obtained using different SiPMs and scintillator crystals of different sizes. We also investigate the feasibility of lowering the electronic jitter by dividing a sensor in segments and using the summation feature of the FastIC, and the possibility of using the ASIC with Cherenkov radiators. We also discuss how the FastIC performance compares with that of other fast-timing ASICs and comment on the feasibility of building clinical ToF-PET scanners with sub-100 ps.
Identifying the photon origin and track on an event-to-event basis is of utmost importance to understand the underlying nature of light based fast timing detectors and to push the boundaries toward the ambitious goal of 10 ps coincidence time resolution (CTR). The photon history can be manifold: different production mechanism (Cherenkov/Scintillation), different materials in a composite structure, variable production point (DOI) in high aspect ratio crystal geometries, photon directionality, or just fluctuations due to the amount of deposited energy. This phenomena shows different classes, timing-wise, of photons, and segregation continues to exist.
This comprehensive study combines three key aspects of exploring the nature of timing with mixed photon origins: experimental work, analytical modeling and image reconstruction.
First, we demonstrate how event classification can be experimentally carried out on the particular case of Cherenkov/Scintillation light with BGO. Timing can be improved by selecting only the fastest 4% of events (CTR: 169 ps → 117 ps FWHM) or “switching off” slow scintillating light and harvesting solely prompt photons, as for instance in PbF2 (142 ps FWHM). Next, an analytical model is developed to estimate/predict the timing performance of either knowing or not knowing the history of the detected photons and to reproduce our laboratory results. Last, measured data on BGO is classified into 25 timing kernels based on the Cherenkov photon yield of each category and used to reconstruct an image of the NEMA IQ phantom, proving from an application point of view the benefits of advanced information.
Although emphasis is given on Cherenkov/Scintillation photons for TOF-PET, this work can be generalized for various types of mixed photon populations and applications, ranging from TOF-CT, TOF-PET, range verification in hadron therapy and high energy physics.
Simultaneous positron emission tomography (PET)/magnetic resonance imaging (MRI) allows the study of molecular and functional processes in the living brain and their spatial and temporal correlation. Our goal is to develop and apply protocols and methods to evaluate molecular changes in receptor and neurotransmitter concentrations by PET and hemodynamic changes by BOLD-fMRI to exploit the complementarity of both modalities for a better understanding of neurological diseases as well as for the development of new treatment strategies. To this end, we are using different rat models and genome engineering technologies targeting specific genes, proteins, and signaling pathways in the brain.
We have heard about outstanding medical research questions/needs, latest tracers and instrumentation developments. Which one is currently the weakest element in (a) the research and (b) clinical context and thus requires special attention?
Where do you see molecular imaging most strongly retain a significant advantage compared to other techniques, for example blood based biomarkers, MRI, etc?
What is your opinion about the utility of preclinical imaging? Where is it most impactful?
The quality of positron emission tomography reconstructed images can be increased by improving the time-of-flight measurement, which is mainly limited by the time response of the standard scintillators (e.g., LYSO, BGO). Another possible option to develop new devices is to use Cherenkov light emitted by electrons moving with velocities exceeding the speed of light in the chosen radiator material. One open question of Cherenkov TOF-PET is how to extend the performance obtained with a small amount of individual PbF2 crystals and SiPMs using modular electronics to crystal and SiPM arrays using a multi-channel readout capable of covering the necessary surface for a clinical whole-body PbF2 TOF-PET scanner. In this work, the feasibility of Cherenkov TOF-PET beyond proof-of-concept experiments using individual crystals was explored. In that sense, the recently developed HRFlexToT electronics was used to experimentally reconstruct images on the scale of two 16 channel Cherenkov TOF-PET modules.
Background: Little progress was made during the last decades to improve the spatial resolution of brain PET scanners, even though the achieved state-of-the-art is far worse than the theoretical limit.
Aim: Develop an ultra-high-resolution (UHR) PET imager based on the field-proven LabPET-II technology platform for human brain imaging at the physical resolution limit.
Method: The UHR scanner relies on truly pixelated detectors featuring 1:1:1 coupling of scintillator, photodetector, and electronic readout to reach ~1.2 mm resolution when imaging the human brain. The basic detector elements consist of 4×8 LYSO arrays of 1.12×1.12×12 mm³ read out by monolithic 4×8 APD arrays, assembled into 128-channel modules with a 2.5D architecture for thermal management and parallel signal processing by custom integrated circuits. The UHR scanner consists of 144 rings of 896 pixels (129,024 detectors) forming a 40-cm diameter by 23.5-cm long cylinder. The useful FOV is up to 30-cm diameter. Single events, along with physiological and motion data, are transferred via Ethernet links to the acquisition computer where a software coincidence engine merges and sorts out list-mode events in real-time for quantitative image reconstruction implementing motion correction.
Results/Conclusions: GATE simulations of the histology based BigBrain atlas demonstrate the ability to resolve FDG radiotracer distribution in cortical or basal brain structures, enabling potential differentiation of details in the entorhinal cortex and locus coeruleus, suspected to be involved in the onset of Alzheimer’s disease. Initial images were obtained using a partial prototype, demonstrating promising results towards microliter volumetric resolution for imaging the human brain.
Dense Cherenkov radiators provide an opportunity for high gamma detection efficiency - due to their high stopping power and photofraction - and excellent coincidence time resolution (CTR). However, because only a few tens of Cherenkov photons follow a gamma interaction in the radiator, the detection efficiency and the energy resolution of a pure Cherenkov detector are an issue. We study gamma detection efficiency and CTR of PbF$_2$ based detectors with different surface treatments and photo-detectors (SiPMs with realistic PDE and 70 ps FWHM SPTR) covering one or two crystal faces (two-sided readout). We investigate the potential performance of a full-size Cherenkov PET scanner using the NEMA NU 2-2018 standard and compare image quality with a reference scanner - Siemens Biograph Vision PET scanner. The geometry of Cherenkov scanners was based on that of the reference scanner. Monte Carlo simulations were performed on a super-computing network using GATE, and CASToR was used for TOF-OSEM image reconstruction. Normalization, scatter, random, and attenuation correction factors were included in the reconstruction. Cherenkov scanner with 1-sided readout had similar TOF performance (225 ps CTR-FWHM) and achieved very similar image quality as the reference scanner. Superior image quality was achieved by using a 2-sided readout detector design (SiPMs at the sides of the crystal), thanks to the improved coincidence detection efficiency and CTR (128 ps FWHM). We demonstrate that even though pure Cherenkov scanners have basically no energy resolution, the scatter fraction of around 50\% is not prohibitively large, and images comparable to the state-of-the-art clinical PET scanner can be achieved due to improved efficiency and CTR attainable with PbF$_2$. Cherenkov detectors are expected to perform even better in preclinical or brain imaging studies where the scatter fraction is significantly smaller compared to the torso imaging.
In time-of-flight positron emission tomography (TOF-PET) excellent coincidence timing resolution (CTR) reduces angular coverage requirements of the scanner, which in turn reduces geometric constraints of the design of the PET scanner. One of the possibilities that this opens is the use of flat panel PET detectors. Such a design would allow a higher degree of modularity and be more compact and cost accessible compared to a conventional ring scanner. Achieving adequate CTR that would allow construction of a flat panel detector that could be used in clinical practice is a considerable challenge and requires improvements at every level of detection. Based on preliminary results from ongoing development of TOF-PET optimized photodetectors and electronics, we expect that CTR of about 75 ps will be achievable on system level in a few years. In this work various designs of flat panel scanners are simulated with point sources and the spatial resolution of resulting reconstructed images is assessed. Simulations were performed using GATE software and image reconstruction was performed using CASToR. The standard methods according to the NEMA standard for determining the spatial resolution, developed for PET scanners with cylindrical geometry, may not be the most suitable for flat panel detectors. A method for flat panel PET scanner spatial resolution estimation is presented. Spatial resolution of simulated flat panel scanners is quantitatively compared to a reference scanner based on current state-of-the-art clinical scanner, Siemens Biograph Vision.
Differentiation between neurodegenerative parkinsonisms, whose early clinical presentation is similar, may be improved with metabolic brain imaging. We identified the characteristic metabolic patterns for Parkinson's disease (PD), multiple system atrophy (MSA) and progressive supranuclear palsy (PSP) – PDRP, MSARP and PSPRP – in a new European cohort.
Brain scans of a Slovenian cohort of 20 PD, 20 MSA-P, 20 PSP patients and 40 healthy controls (HC) were acquired with 18F-fluorodeoxyglucose (FDG) and positron emission tomography (PET). The scaled subprofile model/principal component analysis was applied to identify PDRP, MSARP and PSPRP. Additionally, 56 MSA, 45 PSP, 116 PD and 61 HC subjects were analyzed for validation. We explored the effect of various PET image reconstruction algorithms on the expression of the patterns. We innovatively applied heat-map analysis to extract and graphically display the pattern’s regional sub-scores in individual subjects.
PDRP was characterized by hypermetabolism in pallidum, putamen, thalamus, brain stem, cerebellum and sensory-motor cortex, associated with hypometabolism in posterior parietal, occipital and frontal cortex. MSARP was characterized by hypometabolism in cerebellum and putamen, and PSPRP by hypometabolism in medial prefrontal cortices, nucleus caudatus, frontal cortices and mesencephalon. Patterns’ expression discriminated between PD/MSA/PSP patients and HCs as well as between different parkinsonian cohorts (p<0.001). We confirmed significant correlation of PDRP expression between the reference and other reconstruction algorithms (r≥0.993, p<0.0001). Heat-map analysis showed differences within MSA/PSP subjects and HCs consistent with clinical presentation.
Replication and validation of PDRP, MSARP and PSPRP confirms robustness of these metabolic biomarkers and supports its application in clinical and research practice. We showed that different types of reconstruction algorithms have no significant impact on the pattern expression and that heat-map analysis improves insight into the heterogeneity of studied syndromes by displaying the contribution of various pattern’s regions to patterns’ expression in individual subjects.
With the advancements in the precision of modern radiation therapy techniques and the improved understanding of the underlying biology, there is emerging interest in the integration and harnessing of imaging techniques that can aid the cancer management in the entire treatment chain, from the initial diagnosis through the image-guided treatment up to the follow-up assessment. This talk will particularly review the role of emerging imaging modalities to be integrated in the treatment site in pre-clinical and clinical research, with emphasis on applications in particle therapy, showing the main current achievements along with the remaining challenging demands as well as opportunities for future development.
Computed tomography (CT) is a standard imaging tool for the examination of the head and the entire body. For neurological applications, the detection or rule-out of hemorrhage and stroke are the most crucial questions that have to be answered. The high availability worldwide and the fast acquisition time make CT a valuable tool in an emergency setting.
Dual Energy CT (DECT) provides information on the examined tissue at two energy levels, and therefore offers the possibility to calculate (virtual) monoenergetic images at different energy levels, virtual non-contrast images (VNC), material specific images like e.g. iodine maps and images indicating the atomic number of the scanned materials (z-effective maps).
The aim of this presentation is to show several clinical scenarios where Dual Energy CT and Photon Counting CT can improve diagnosis in neurological applications
The paradigm shift in medicine from treatment of acute and/or advanced disease to very early diagnosis and even prevention in cancer, neurodegenerative as well as cardiac fields puts more stringent requirements on PET imaging both in terms of sensitivity as well as specificity. Likewise, recent developments in Targeted Radionuclide Therapy (TRT) where theragnostic pairs are used to tailor a personalized treatment in terms of dose using PET initial imaging and subsequent alpha or beta emitting radionuclides have introduced a clear and urgent need for more widespread and accurate PET imaging. Standard clinical scanners are sub-optimal both in terms of cost that limit widespread use as well as performance. Standard clinical PET scanners use sets of tightly arranged rings of detector modules, usually consisting of scintillation crystals optically coupled to light sensors with readout electronics, cover only a limited solid angle, and just a small few percent fraction of the positron decays is registered.
In this talk, we present a clinical overview of the need for high sensitivity and/or high resolution scanners in specific clinical scenarios.
Two specific scenarios at the two ends of the spectrum will be covered: dedicated brain PET with ultra-high spatial resolution in the context of neurodegerative and neurological diseases; and long axial high sensitivity very fast timing whole body PET in the context of cancer. Designs and preliminary results of the two systems will be presented and potential clinical significance discussed.
Spatial resolution is a key factor in preclinical and neuro-PET imaging. By minimizing the partial volume effect, and the resulting spillover from/to surrounding tissues, the accuracy for quantifying the radiotracer concentration in small structures can be significantly enhanced, therefore mitigating the need to maximize the sensitivity for increasing the contrast-to-noise ratio in target tissue. The technology to image small animals in pre-clinical studies has been developed and improved over the years to achieve sub-mm (or sub-µL) spatial resolution, close to the physical limit imposed by the positron range and the annihilation photon acollinearity. Recent dedicated small animal PET scanners, such as the LabPET II, now enable the quantification of regional radiotracer concentration in the brain and other organs of rats and mice. However, little progress was made during the last two decades to improve the spatial resolution of dedicated brain PET scanners, even though the achieved state-of-the-art is far worse than the theoretical limit. While the detection technology developed for pre-clinical scanners would allow the theoretical limit to be reached in brain PET, it is generally not adapted and too costly for larger scanner designs. With the improvement of manufacturing processes, the large scale integration of front-end electronics and the development of a modular and scalable detector design, it is now possible to implement the LabPET II detector technology in a dedicated ultra-high-resolution (UHR) brain PET scanner approaching the theoretical physical limit of resolution at 1.25 mm FWHM (~2 µL). The initial imaging performance of the partial UHR scanner will be presented.
PET imaging is essential in nearly all aspects of high-quality cancer care. However, challenges to the delivery of PET radiopharmaceuticals in developing countries are manifold. Developing countries typically face shortages of imaging equipment and nuclear medicine physicians, and efforts to improve PET imaging are complicated by poor infrastructure, economic barriers, and other obstacles. In addition, financial issues are essential in the implementation and maintenance of producer centers, especially for developing countries. This talk is intended to raise awareness of some of these challenges by exploring the Brazilian example. The inequities in the availability and affordability of PET imaging facilities among different geographic regions will be discussed, as well as efforts and strategies to address these shortages.
How to build systems that matter? To this, we made some practical observations in the narrow field of brain PET in the last years. The applications of brain PET in diagnostics for primary brain tumors, dementia, movement disorders, epilepsy, drug development, and neuroimaging research were so far not enough to warrant a revival of dedicated brain PET systems. The anticipated use together with treatments against Alzheimer’s disease is still lacking, while noninvasive tau and amyloid diagnostic competition from plasma-assays is rising.
Yet, in this talk, we point out three macro-observations. First, in larger PET-centers, the upcoming whole-body PET/CT systems are raising the price pressure, favoring high throughput routine oncology applications. This has an adverse effect on neuroimaging in these centers. Second, in case of the anticipated application of Alzheimer’s disease treatment accompanying diagnostic, the need is so large, that plasma-assays and PET imaging will likely co-exist. Third, brain PET applications and reimbursement practices are growing, also besides Alzheimer’s disease treatment accompanying diagnostic, which might become market drivers on their own.
We will share our conclusions from these observations and from clinician’s interviews that have guided our system development of the dedicated brain PET system NeuroLF. In the end, the bar is to cross the threshold of relevance: The frequency of procedures at which a dedicated system becomes economic to use.
We extracted a core vision, in which NeuroLF addresses challenges that clinicians and clinical researchers face already today. We point to our system optimizations in improving system performance that matters in an efficient way, in the untapped potential in open brain PET data and open software communities, in stepping towards dynamic imaging in the clinical routine, and in bringing PET to the patient with small-scale ambulatory and point-of-care systems.
Full brain imaging in humans during natural body movements is challenging because most high-resolution imaging modalities (PET, MRI, CT) require participants to lie down motionless, limiting the opportunity to study upright locomotion to that of imagined movement. Because PET imaging involves the injection of a radiotracer for imaging, it has been possible to image full brain activation during movement by imaging tracer accumulation that occurred during continuous locomotion, for example. This type of imaging is very informative as it provides a picture of the brain regions involved in a continuous task that occurred during radiotracer uptake (~ 20 minutes). However, it does not provide temporal or event-related information, crucial to understand the sequence of brain activations necessary to control our movements. I will discuss recent developments in dynamic radiotracer administration, integration of multimodality imaging that could enable the exploration of the neural control of the human during whole-body movements. This potential paradigm-changing advance in brain imaging capacity would be a turning point for fundamental and clinical studies in understanding basic mechanisms, training- or disease-induced changes in brain function during movement. The development of such applications could extend to event-related changes in drug therapy or disease-related symptoms.
Real-time dynamic positron emission tomography (PET) of human brain function in natural and free states has significant and far-reaching scientific and clinical value. However, PET detectors need to use scintillation crystals with high effective atomic number and high density to detect gamma photon pairs generated by the positron annihilation, which is difficult to meet the requirements of light weight and miniaturization in the design of wearable imaging systems.
Advanced Time of Flight (TOF) technology can reduce the uncertainty in the estimation of the position of positron annihilation during PET image reconstruction, improve the effective sensitivity of the system and the signal-to-noise ratio of the image, and significantly reduce the usage and weight of the scintillation crystals. Therefore, TOF technology is a key technology for realizing wearable and mobile PET imaging.
We are developing a TOF-PET system for wearable and mobile brain imaging. The wearable PET helmet (118 mm in radius and 134 mm in length) is constructed with 2.7kg of LYSO crystals. In this talk, we will report our progress in the simulation-based system design, the designs and performance of the TOF PET detectors, readout electronics, and mechanics, and the performance of the assembled system.
PET human brain imaging has evolved dramatically, using specific radiotracers and imaging paradigms to measure numerous brain targets and to assess neurotransmitter and receptor dynamics. Until recently, dedicated brain PET has not progressed since the HRRT, so there is a compelling need to build next generation human brain PET systems. This is the goal of the NeuroEXPLORER (NX) project. Based on experience with >5000 human brain HRRT studies at Yale and with the total-body uEXPLORER system at UC Davis, the NX is being constructed to meet the following design goals: 1) Ultra-high sensitivity, to be achieved with a long axial field-of-view (aFOV) plus excellent time-of-flight (TOF); 2) Exceptional image resolution through small detectors, depth of interaction (DOI) readout, and corrections for inter-crystal scatter; 3) Continuous head motion tracking and correction. The NX design is a cylinder with diameter of 51.6 cm and aFOV of 48.1 cm. The system consists of 5 complete detector rings, with an additional incomplete 6th ring to accommodate shoulders to place the brain in the aFOV center. LYSO crystals (20-mm deep) have an in-plane dimension of 1.5 mm, leading to a resolution of 1.6-1.8 mm. To reduce parallax error, a novel single-end DOI design was developed with < 4 mm FWHM. The projected TOF resolution is < 270 ps. Combined with the long aFOV, we project that the NX will have > 10-fold higher effective sensitivity than the HRRT. Head motion tracking is performed with a real-time stereovision system. Optimization of reconstruction and quantification is performed with high-resolution brain simulations and novel phantom configurations. Ultimately, human paradigms will demonstrate the effectiveness of the NX: 1) showing the dramatic sensitivity increase compared to the HRRT, 2) leveraging high sensitivity to reliably measure uptake in small nuclei, and 3) opening new frontiers of imaging neurotransmitter dynamics.
The PETITION (PET for InTensive Care units and Innovative protON therapy) collaboration is currently developing two compact brain Positron Emission Tomography (PET) scanners with a focus on new clinical applications. The first device (ICU device) is a mobile scanner intended to scan deeply sedated, critically-ill patients with sepsis directly in the Intensive Care Units (ICU). The second device (PBT device) is designed for proton beam therapy (PBT) for the purpose of online in vivo dosimetry and hypoxia tracking studies.
The PET scanners are built with custom-made mechanics to match the different requirements of the two applications. While the ICU device is a conventional full ring scanner, the PBT system features an opening which allows for proton irradiation during image acquisition. The detector readout electronics are identical for both system. Lutetium-yttrium oxyorthosilicate scintillators arranged in blocks of crystals are combined with Silicon Photomultipliers as active detector elements.
The image reconstruction software is based on the well-known ordered subset expectation maximisation (OSEM) algorithm. For quantitative PET, image corrections are imperative and commonly external information is used for scatter and attenuation estimates. However the PETITION devices have neither radionuclide transmission sources nor X-ray computed tomography available, and emission only data correction has to be used. While the transmission-less method is already used with full ring PET scanners, it is usually not used with open ring devices.
This document gives a quick overview of the PETITION scanners before presenting preliminary results of image reconstruction and correction based on emission data only for both PETITION scanners geometries.
Neurodegenerative dementias are a group of slowly progressing neurological disorders with the most common being dementia due to Alzheimer’s disease (AD) and dementia with Lewy bodies (DLB). Despite different main pathology and symptoms, there can be a substantial overlap in concomitant pathology and clinical presentation and misdiagnosis is not uncommon even at specialized dementia clinics. Metabolic brain imaging with 2-[18F]fluoro-2-deoxy-D-glucose positron emission tomography (FDG PET) is an imaging technique that provides in vivo information about synaptic (dys)function. The latter arises due to accumulation of abnormally folded proteins and precedes neuronal death. Application of multivariate approaches to FDG PET analysis have enabled identification and validation of disease-specific metabolic patterns, which expression can be prospectively quantified. Using clinical and imaging data from over 500 participants from different centres, we identified, validated and explored diagnostic and prognostic values of disease-related metabolic brain patterns. AD-related pattern (ADRP) is characterized by reduced metabolic activity in temporoparietal regions and precuneus with co-varying increased metabolic activity in the cerebellum, pons and sensorimotor cortex. Pattern expression scores can differentiate between AD and healthy controls, they correlate with measures of cognitive impairment, increase linearly with time and predict progression from mild cognitive impairment to dementia. DLB-related pattern (DLBRP) is characterized by reduced metabolic activity in occipital, parietal and temporal cortices and precuneus with co-varying increased metabolic activity in basal ganglia, medial temporal lobe, cerebellum and pons. Pattern expression scores can differentiate between DLB and healthy controls, they correlate with measures of cognitive impairment and survival time. Furthermore, after accounting for topographic overlap between ADRP and DLBRP, we can accurately differentiate between AD and DLB patients. Lastly, we recently showed that a differential algorithm based on pattern expression scores and machine learning can achieve similar sensitivity and specificity to visual reading of an expert clinician.
Alzheimer’s disease (AD) is the most common neurodegenerative brain disorder and AD-related metabolic pattern (ADRP) is its biomarker that has shown promising results in several studies. One of the obstacles towards successful translation to the clinic is limited evidence on its diagnostic performance with incoherent imaging data - for example different image resolution and identification group sizes.
In total 240 FDG PET images (120 AD/120 cognitive normal (CN) subjects) were selected from the Alzheimer disease neuroimaging initiative (ADNI) database. Out of that, 200 images (100 AD/100 CN) were used to identify ADRPs with scaled subprofile model/principal component analysis and were randomly-selected 25 times for each of five different identification group sizes (20 AD/20 CN, 30 AD/30 CN, 40 AD/40 CN, 60 AD/60 CN and 80 AD/80 CN) and six different image resolutions (6, 8, 10, 12, 15 and 20 mm). In total 750 ADRPs were identified and validated through AUC values on remaining 20 AD/20 CN group with six different resolutions.
We found that small identification cohorts (20 AD/20 CN) may be adequate in some cases, but these results are up to 0.3 AUC different for groups with different subjects. Slightly larger identification cohort (30 AD/30 CN) gives far more consistent results (0.2 AUC), and further increase improves the independence of subject selection even more (0.1 AUC). Image resolution is not a major factor in successful ADRP identification, as long it stays within a loosely set range of 10-15 mm. Match in resolution for identification and validation images does not affect the ADRP diagnostic performance.
While small identification cohorts might be adequate in good subject’s selection, larger identification cohorts are preferred to avoid unfavorable selection of identification subjects and consequent compromised ADRP diagnostic performance. Properly identified ADRP can be used on data from different scanners with different image resolution.
Introduction: Multivariate metabolic brain patterns obtained by SSM/PCA analysis of [18F]FDG-PET scans represent discrete functional brain networks. They were identified and validated in most neurodegenerative dementias and are routinely used in clinical practice. However, the internal structure of these networks remains unknown. In this study, we explored metabolic connectivity with graph theory methods within the predefined SSM/PCA derived networks in two neurodegenerative dementias: Creutzfeldt-Jakob’s disease (CJD) (in vivo model of neurodegeneration due to fast progression/pathological availability) and frontotemporal dementia (FTD) (for robust and consistent network).
Methods: Topographic maps of CJD-related (CJDRP) and FTD-related patterns (FTDRP) were transformed to 95 regions-of-interest. Regions with standardized weights above and below one standard deviation were defined as disease specific nodes. For each node, normalized metabolic activity was calculated for corresponding patients and normal controls (NC). Metabolic data were used to construct matrices of node-to-node pairwise correlations separately for patients and NC. Bootstraping (100 iterations) was used to estimates correlation pairs. Global connectivity differences between patients and NC networks in both spaces were studied by degree centrality, clustering coefficient, characteristic path length, small-worldness, and assortativity at varying graph thresholds for all the bootstrap iterations.
Results: In both vectors spaces (CJDRP and FTDRP) we observed signifficant connectivity changes. While degree centrality was elevated and clustering coefficient/small worldness were decreased in CJDRP space in CJD patients compared to NC, it was the opposite in the FTDRP space/FTD patients. However, path length and assortativity were elevated in both diseases showing reduced efficiency of information transfer. In CJDRP, we observed disruption of normal connections with compensatory reconnections, while only severe disrupting was observed in FTDRP.
Conclusions: Functional connectivity exploration within disease-specific network spaces enable us to understand network’s internal organization. Connectivity measure changes are not constant across different disorders and should be understood in context of network disruption and adaptations.
The PET-CT technology and its continuous developments are widely spread all over the world, however the distribution of the cyclotrons and the availability and affordability of its tracers is not well organized among the developing countries and varies greatly from country to another and consequently will affect the number of installed PET-CT scanners. According to the IAEA database the number of PET-CT in the middle east region is one scanner per 2.04 million people. Recommendations provided in literature suggests that the number of PET-CT scanners should be 1.0-1.5 per million inhabitants. In the African continent there is a great heterogeneity related with nuclear medicine and PET-CT equipment available, with some countries having the highest technology including cyclotrons and (PET/CT), and other with no nuclear medicine units. Seventeen cyclotrons supply 63 PET/CT scanners within this` continent. Six countries only out of 52 concentrate all cyclotrons facilities, most of them are managed by private companies. Several challenges and obstacles are facing the proper availability of such technology among developing countries and in consequence the proper healthcare delivery for patients. Those obstacles include political, regulatory, economical, infrastructure, marketing and manpower problems. On the other hand, finding solutions for those obstacles is applicable as there are several alternatives and methodologies has been applied to overcome those obstacles in the developing countries.
The impact of AI methodology vs dedicated scanners (or a combination of both) on research and clinical application
Positron Emission Tomography (PET) is one the most used techniques in Medical Imaging. It originated from the synergic discovery of 5 scientists, all Nobel Laureates in three different fields of science (Physics, Chemistry, Physiology or Medicine), namely: Carl David Anderson (Nobel Laureate in Physics in 1936) “for his discovery of the positron”; Ernest Orlando Lawrence (Nobel Laureate in Physics in 1939) “for the invention and development of the cyclotron and for results obtained with it, especially with regard to artificial radioactive elements”; George de Hevesy (Nobel laureate in Chemistry in 1943 ) “for his work on the use of isotopes as tracers in the study of chemical processes”; Godfrey N. Hounsfield and Allan M. Cormack, (Nobel Laureates in Physiology or Medicine in 1979) “for the development of computer assisted tomography”, that uses the same image reconstruction method as utilized by PET. It is notable that the first concept of a PET scanner was proposed by a neurosurgeon (William Sweet) as a device to localize a tumor in the left or right part of the brain. The preliminary idea was presented at the dedication of the Research Building of the Massachusetts General Hospital on May 16, 1951. More than 70 years have passed since then and a “plethora” of applications have been done and PET is an essential actor in the diagnosis, therapy and prognosis of many diseases.
This list of Nobel laureates outlines the 4 pillars of PET: the physics of the positron, the production of artificial radioisotopes, the use of the radiotracers to study the physiology and pathology of the human body, the 2-D, 3-D and 4-D reconstruction of the activity image within an organ, a system or the entire body. In this presentation, I will make an “on the flight” review of many applications, and some thoughts will be presented for the future.
I will try to cover the numerous PET applications from the initial studies of the physiology of brain, to tumor detection and tumor biology, to the contribution to treatment planning and particle therapy range monitoring. The evolution of the detectors (wire chambers, high Z and ultra-fast pixelated scintillators), phototubes and solid state photosensors, discrete and digital electronics, MLEM reconstruction algorithms and, more recently, Artificial Intelligences have well established the field of Molecular Imaging. The solid angle coverage of a PET scanner has already increased to almost 4 Pi, whereas Time-of -Flight PET (TOFPET) aims reaching 10 ps resolution.
The future for PET is challenging and promising at the same time. Further improvements are behind the corner: for instance, to couple PET to metabolic and metabolomic radiotracers; new information will be provided from hybrid system such as PET/MR or EEG/PET/MR coupled to AI, where the immediate “ms” response to electric stimuli will be combined with the “some s” response of fMRI and with the “several minutes” stationary information of PET. The split-gradient MR, already available for assisted MR radiotherapy, could provide the modern hardware scenario. Organ specific PET, especially for Breast and Brain, may become available at a reasonable cost and additional positron physics could be used such as Cherenkov imaging or three photon detection for appropriate radioisotopes. All in all, after 70 years on the scene PET is alive and kicking and will be a fundamental and truly interdisciplinary imaging technique of the “precision medicine” for years to come.