Preview

PDF, English Download (202MB) | Terms of use

Abstract

Efforts towards the development of a detector system designed for simultaneous optical and nuclear imaging, i.e. combining fluorescence/bioluminescence techniques with PET/SPECT modalities, have started over a decade ago. Given the potential that would be introduced by the capacity of simultaneously detecting gamma-rays and optical- probe photons, a concept for an integrated system being capable of achieving such task was developed in this work. The blueprint for an integrated preclinical PET-OT system consisting of detector modules that combine monolithic scintillating crystals with plenoptic detectors is presented. Each module of this system is specifically composed of a 25x25x25mm LYSO crystal with two 8x8-channel photosensors of matching 2D size coupled to its back face and to another lateral face of the crystal in a perpendicular dual- readout configuration. The plenoptic detectors consist of a CMOS flat sensor of corresponding 2D size with a microlens array attached to them. A simulation and reconstruction framework for the investigation of monolithic-detector PET systems was developed for the conception and analysis of the proposed system. Within this framework, two alternative algorithms required for the operation of monolithic-detector PET systems were developed respectively based on a k-nearest neighbours and a convolutional neural network approach. The POI estimation FWHM-resolution of the former was found to be 0.68mm, 0.66mm and 0.52mm in the x, y and z dimensions respectively for the monolithic-detector module in the proposed system. The latter presented a POI estimation FWHM-resolution of 0.42mm, 0.41mm and 0.49mm in the x, y and z dimensions respectively. The crystal depth, the number of crystal faces to which a photosensor is coupled and the number of channels per photosensor were optimised in order to reach the final proposed design.