Sensitivity in PET: alternate method for Compton scatter discrimination
based on neural networks
por
Jean-Baptiste Michaud
(University of Sherbrooke)
→
Europe/Madrid
Sala Seminarios IFIC (Edf. Institutos de Investigación)
Sala Seminarios IFIC
Edf. Institutos de Investigación
Descripción
Sensitivity in PET can be increased by lowering the energy threshold, to include Compton-scattered photons. However, exact and correct analysis of those photons is statistically uncertain, especially within a high-resolution scanner with a large number of noisy detectors. This talk presents the work-to-date on an alternate method that is based on neural networks. A proof of concept has been conducted, as well as GATE Monte-Carlo simulations. Results to date show significant improvement over "traditional" methods based on explicit Compton-kinematics analysis.