Tomographic reconstruction for 3-photon imaging based on hybrid PET/Compton cameras
Abstract
Nuclear medical imaging covers several functional imaging modalities based on the injection of a radiotracer targeting a specific process in the body, the detection of the emitted radiation, and the tomographic reconstruction of the detected signals to obtain an image of the spatial distribution of the radiotracer. Recent instrumental developments combine two of these modalities: positron emission tomography (PET) and Compton imaging. These hybrid PET/Compton systems can detect PET and Compton events independently, as well as so-called 3γ events, taking advantage of both modalities. In this work, we contribute to the field of tomographic reconstruction for these hybrid systems. Reconstruction requires a direct model linking the image to the events, called a projector. We propose an original Compton projector method that improves the trade-off between precision and computational cost compared to standard methods. We propose a reconstruction framework combining the statistical contributions of each type of event. The developed algorithm makes it possible to associate PET or Compton events alone with 3γ events, generally detected in smaller quantities, thus improving the estimation of the activity distribution. We apply these developments to XEMIS2, a monolithic liquid xenon Compton camera dedicated to 3γ, currently being installed at the Nantes University Hospital.
