Author: Yuemeng Feng, Hamid Sabet 👨🔬
Affiliation: Massachusetts General Hospital, Harvard Medical School 🌍
Purpose: We propose a novel brain-dedicated PET system comprising elliptical cylinder with a neck cut-out, supplemented by front and back panels to improve sensitivity and line-of-response sampling. The design concept aims to maximize solid angle coverage of the human head.
Methods: The proposed system comprises a total of 526 LYSO: Ce detector modules with each module measuring 25x25x20 mm³. Each layer of the detector is designed to be coupled one-to-one to Hamamatsu SiPM array with 1.5 mm pixel pitch. The scanner consists of 12 elliptical cylinders, one back panel, and one front panel. The cylindrical part includes 4 fully populated rings (4x34 modules) and 8 partially populated rings (8x25 modules) to accommodate the patient's neck. The back panel contains 102 modules, and the front panel has 88 modules with a 5x10 cm² opening for virtual reality camera and motion tracker. This work presents resolution-sensitivity metrics using detector configurations with varying DOI levels, employing a 20 mm thick detector and timing resolution values varying from 200 ps to 50 ps.
Results: The reconstructed images of a Derenzo-likephantom and an extended cardiac and torso, and brain voxelaized sources are presented. Results show that the sensitivity varies from 35.04% to 13.59% across the FOV. He lowest sensitivity is in the carotid artery region which is much higher than other brain PET systems. The reconstructed image resolution varies from 1.0 mm to 1.5 mm within the FOV without applying any resolution recovery methods.
Conclusion: The proposed brain-dedicated PET system shows promising spatial resolution and sensitivity far exceeding alternative solutions. The modelled detector and its pixel dimensions are specifically chosen to represent available photodetector and readout technologies for practicality reasons. We are exploring multiple options to achieve the required DOI level, including fast waveform digitizing ASICs with built-in DOI measurement using light reflection ratio.