Optimized reconstruction of the position of interaction in high-performances γ-cameras

Vectorized internal radiotherapy is an efficient modality for cancer treatment but requires a personalized dosimetry to adapt the administered dose for each patient, in order to limit the toxicity to organs-at-risk and maximize therapeutic effects. This can be done by performing quantitative imaging of the target organ with high resolution γ-imaging devices in order to evaluate the uptake and biokinetics of the radiotracer. We developed a high-resolution portable γ-camera with a 10 × 10 cm2 field of view dedicated to thyroid imaging during diseases treatments with 131I. In addition to the optimization of the detection elements, the quality of the images also depends on the implementation of efficient methods to reconstruct the interaction position of gamma rays in the monolithic scintillator from the measurement of the scintillation light distribution. We present here the results obtained with two different machine learning methods based on experimental data for reconstruction of the γ-rays interaction position. Those methods reach high spatial performances such as millimeter spatial resolution and submillimeter distortion in the center of the field of view.