Feasibility study of a concurrent image reconstruction algorithm for proton therapy with in-beam TOF-PET

A real-time dose-guidance is highly desirable for monitoring the accuracy of the radiation treatment during irradiation. We present a sequentially processed image reconstruction algorithm (SPIRA) for in-beam TOF-PET that has the potential to provide concurrent imaging of the positron distributions from list-mode TOF-PET data in particle therapy. The SPIRA employs the maximum-likelihood (ML) criterion and reconstructs an image in the continuous-coordinate event space. In parallel, a serial processor (SP) is adopted to support concurrent imaging. The algorithm has been applied to computer simulation data generated for its potential use in in-beam TOF-PET monitoring of dose distributions in proton therapy, including a full-ring system and a dual-head partial-ring system. We considered various coincidences resolving times (CRTs) for the systems including 300 ps and 600 ps. Two opening angles Ø, 45 and 90°, were considered in the partial-ring geometry. The images generated by the SPIRA appear to provide better resolution and contrast recovery than by the conventional image reconstruction method in which the events are simply placed at the computed positions based on the TOF measurement. We believe that SPIRA can be useful for in vivo treatment verification in a real-time dose-guided particle therapy.