Abstract
317
Objectives: It has been shown that radio-guided probes can be used to identify sentinel lymph node and to delineate tumor which ultimately lead to improved patient recovery in cancer patients. However, no single-modality probe (gamma-, positron-, or NIR-based) has been able to achieve less than %5 false negative rate (FNR), a key metric suggested by American Society of Clinical Oncology. In this regard, we are developing a feature-rich multi-modality intraoperative imaging platform to be used with positron emitting isotopes such as 18F-FDG for surgical applications. A key component of this platform is PET-like detector system consisting of a high-resolution standalone detector (SD) with time-of-flight (TOF) capabilty, in coincidence with a half-ring detector system (HDS) in that the SD is placed in various locations wrt field of view (FOV) while HDS is fixed underneath the patient bed. This falls into category of limited angle tomography, with the aim of accurate estimation of lesion depth and boundary during surgery where the use of a complete PET system is not feasible.
Methods: We modeled the intraoperative system and a reference PET using monte-carlo code GATE. Geometry 1: As a reference, we modeled a 3-ring whole-body (WB) PET with 44 detector blocks per ring. Each 5x5 cm2 detector block consists of 13x13 array of 3.6x3.6x20 mm3 LYSO:Ce pixels with 3.8 mm pitch. Geometry 2: We modeled a system similar to Case 1, but with only 1 (or 3, or 5) detector blocks in coincidence with the bottom half-ring detectors. Geometry 3: a SD placed in proximity of the FOV moving laterally in coincidence with detectors in the bottom HDS. We are investigating how TOF, fast coincidence timing resolution (CTR), and location of SD can improve image quality. Image quality was evaluated using a phantom consisting of radioactive spheres ranging from 1-5 mm diameter placed in L-shape with 2 cm distance between center of spheres. Same L-shape arrangement of hot spheres were repeated in axial directions at 3 and 6 cm from the center. Activity concentration was kept constant across all hot spheres at 238 bq/milliliter. Hot spheres were placed inside a 16 cm elliptical water tube with 25 and 45 cm short and long axes, respectively. Ordered subset expectation maximization (OSEM) with 3 iterations and 4 subsets was used for image reconstruction. Images were reconstructed with no TOF information as well as for CTR values of 150, 250, 350, 450, and 550 ps. SD and HDS detectors have the same intrinsic resolution in this study.
Results: Geometry 1: Results show that with a 2-minute acquisition and full ring detector blocks, one can clearly resolve 3-5 mm spheres but as expected, 2 mm sphere is only marginally resolved when TOF with 150 ps CTR is used. In Geometry 2, as expected the image is improved as we increase the number of detector blocks from 1 to 3, and to 5 in the upper half-ring, and further improved when short CTR is implemented. In Geometry 3, we observed that placing the SD closer to the FOV compared with Geometry 2, improves the image quality. Even with 3 SD positions and CTR of 150, one can partially resolve 3-5 mm hot spheres.
Conclusions: The results suggest that to improve the image quality with minimum number of SD positions, detectors with fast CTR such as 150 ps is needed. A combination of larger solid angle in geometry 3 and high CTR are key to successfully resolve lesions with 5 mm diameter. It is apparent that utilizing SD and HDS detectors with better intrinsic resolution is necessary to resolve smaller lesions.