Adding Position-Sensitive Functionality to a Forceps-Type Mini-PET Prototype

Optimizing the extent of lymph node (LN) dissection in surgical cancer treatment is expected to minimize the postoperative complication rate. Our previous clinical study reported that
¹⁸
F-fluoro-2-deoxy-D-glucose (FDG) could be a biomarker for intraoperative LN metastasis diagnosis. To realize an intraoperative LN diagnosis with FDG, we are developing a forceps-type mini-PET as an intraoperative device. With a pair of small scintillation detectors mounted on the forceps head, the device can detect annihilation radiations from FDG by pinching a target LN individually. However, the radiation-detection efficiency strongly depends on the position of the radioactivity source in the field-of-view (FOV) between paired detectors, which may cause false judgment in LN diagnosis. To correct the efficiency variation, the positional information of the radioactivity distribution is necessary. Therefore, in this work, we proposed to add a position-sensitive functionality to the forceps-type mini PET. The basic concept is to divide each monolithic scintillator into a 2×2 matrix (consisting of four crystals). The center position of the 18F distribution can be estimated from 4×4 signals by applying a simple neural network (NN), which was proven in our previous simulation study. In this study, we developed the prototype device having the 2×2 matrix and evaluated the source position estimation performance. The NN was tuned by using Geant4 simulation data. The energy resolution was 27 % at 511keV, and the coincidence timing resolution was 9.24 ns. The proposed method for source position estimation using the NN worked well, and the accuracy of the z position was within the error of 2 mm.