Hadron Therapy Range Verification via Machine-Learning Aided Prompt-Gamma Imaging

The aim of this work is to demonstrate the capability of the i-TED Compton imager combined with ML-techniques for enhanced performance range verification in prompt gamma-ray monitoring. Prompt Gamma monitoring constitutes a promising technique for range verification in hadron therapy treatments. Hadron Therapy with protons introduces advantages with respect to the conventional radiotherapy because of the maximization of the energy deposition (dose) at the Bragg peak. i-TED is an advanced array of Compton cameras originally designed for neutron-capture nuclear experiments. However, due to its large detection efficiency, fast response, high time resolution, compactness, low sensitivity to neutron-induced backgrounds and image resolution, i-TED shows also an excellent performance for medical purposes such as PG monitoring. Furthermore, aiming at improved quality Compton images in the high-energy gamma-ray range characteristic of Hadron Therapy, a novel Machine Learning methodology has been developed and applied for identification of full-energy events. To that purpose, a detailed GEANT4 Monte Carlo study simulating a clinical irradiation has been performed. Finally, we present first results of the novel Machine Learning methodology applied to an experiment performed at the cyclotron of the CNA facility, Spain, where two i-TED modules were simultaneously operated as in-beam PET and Compton imagers.