Feasibility study of imaging with tissue-scattered triple-γcoincidence events in Compton-PET

The objective of the present work is to study Compton-PET imaging with single scattered (in tissue) data using a non-pure beta emitter (44Sc). Such radioisotopes are advantageous because of the extra gamma radiation to be used in triple- 3 coincidences. Utilizing the single scattered events in conjunction with an additional gamma (1157 keV) emitted from 44Sc, provided us with an opportunity to obtain the direct images of the source activity distribution. The idea presented here is unique because there is hardly any imaging research with tissue-scattered data in the literature, especially for a Compton-PET system using a non-pure beta emitter. Initially, we described the locus of tissue-scattered point as a prolate spheroid surface or spindle toroid, which became two intersection points after using Compton cones from both ends of 511 keV photon detections. We considered a human-sized Compton-PET system of two types - (i) ideal and (ii) state-of-art scanner - depending on the energy and timing resolutions of detectors. The list-mode data were generated via GATE simulations. The appropriate trigger logic was applied using the two-window method to separate the events corresponding to single tissue-scattered and true (unscattered) coincidences. The imaging involves estimating the annihilation point for each triple-coincidence event using Compton cone of 1157 keV gamma and time-of-flight information for the 511 keV. Thus, we directly obtained images of the activity distribution without using any reconstruction algorithm. They turned out indeed meaningful, establishing our feasibility study. For quality assessment, we compared the full-width-half-maximum (FWHM) of the intensity profiles of various images. For ideal scanner, for instance, the FWHM for the tissue-scattered image was almost double of that for the unscattered one. Although the image quality was not all that good, but we expected it because of physics-based uncertainties involved in the tissue-scattered imaging. Nonetheless, separate assessments of the underlying activity from two independent data sets - one from single tissue-scattered, another from the usual true coincidences - could certainly improve image interpretation in medical diagnosis.