Speaker
Description
PET scanners are indispensable imaging tools in modern medicine. In recent years it was shown that a planar configuration, in contrast to the classical cylindrical shape, could become a practical option, while capable to operate in challenging environments and protocols, such as surgery suit, ER, upright imaging, that are not possible using current PET scanners. For such an open tomograph geometry to work, measurement of the positron annihilation into 2 photons must be defined with TOF accuracy better than ~100ps FWHM. We present results of a small prototype system studies, used to investigate optimal configurations scalable to larger detectors.
Summary (500 words)
PET scanners are of great importance in health care, with an aging population in EU and worldwide. In recent years access to diagnostics has improved, yet still remains on a low level of a few machines per million inhabitants. A robust planar PET scanner is appealing in clinical use from multiple perspectives. For example, as a mobile, it can come to the bedside, it doesn’t require a moving bed, it can operate in tight environments such as ER or ICU, a patient can be seated during the procedure, it is less claustrophobic, etc. For such an open geometry to work, a coincidence time resolution (CTR) below 100ps is needed, to recover the image quality and contrast [1-2]. It was demonstrated with the high frequency (HF) amplifier that using small (2x2x3mm) LYSO or LSO crystals this can be achieved [3]. However, the approaches investigated so far were not scalable.
We present findings from a two-fold investigation. First, using the HF amplifiers setup (Fig.1a) based on BGA 616 amplifiers for timing, and AD8000 for energy, paired with FBK NUV-HD-MT low field, 3 mm x 3 mm, 40 um LF-M0, we categorized CTR FWHM obtained from various sizes and crystal producers.
The CTR results in figure 1b were obtained using the LYSO 2x2x3mm, at different sampling rates, and low-pass filtering settings. This research aims to define an optimal crystal size, timing vs detection efficiency balance, while signal filtering aims at understanding the necessary front-end bandwidth to achieve our goals. Figure 1b shows that oversampling doesn’t provide any benefit, and it even degrades the result.
Second, we built a small mockup consisting of 4 boards, each equipped with two FASTic ASICs [4] as shown in figure 2a. SiPMs from FBK, the NUV-HD 4x4mm, 40um, M0 and the NU-HD-MT 3x3mm, 50um. They have separate bias RC filters, and are connected to the readout ASICs in both polarity for optimal coupling exploration. Boards can be programmed, and calibrated via an FPGA, while the FASTic outputs are connected to the CAEN PicoTDC module [5]. With this setup we aim to understand the challenges to overcome towards a truly scalable planar PET module. Having individual tests set to measure sensitivity to single photon detection, as shown in figure 2b, SPTR and signal headroom provides insightful information for fine-tuning. In summary, these currently ongoing experiments and simulations and the development of advanced fast ASICs, will benefit the PETVision collaboration [6] working on a scalable ultra-fast planar TOFPET tomograph. And importantly, these scanners are expected to be less expensive than the current scanners, by using much less of the very expensive detector material.
