Modeling of high resolution depth of interaction detector for small animal in-beam PET scanner

The development of novel PET imaging systems plays a key role in advancing in-beam range verification techniques for particle therapy. Within the ERC-funded SIRMIO project at LMU (www.lmu.de/sirmio), a compact, high-resolution PET scanner has been developed to accurately monitor dose delivery in preclinical proton irradiation studies. This thesis contributes to optimizing this technology by modeling and characterizing the depth-of-interaction (DOI) detector response of a single detection module, utilizing both experimental data and Geant4-based Monte Carlo simulations. Experimental measurements were performed using a micro Derenzo phantom (MediLumine, Canada) filled with 18F-FDG, supplemented by acquisitions with a 22Na source. A dedicated analysis pipeline, combining ROOT frameworks and custom Python scripts, was developed to extract precise spatial and spectral response parameters for each detector pixel. Parameters analyzed include pixel positions, spatial spread, and per-channel energy distributions. The generated pixel-wise database provides detailed insights into the detector’s response, significantly advancing the understanding of spectral and spatial features at the pixel level. Although this pixel-wise model has not yet been implemented for event-level filtering, it represents a critical step toward advanced classification methods aimed at selectively retaining true 511 keV single-interaction events. Ultimately, this foundational modeling work is expected to improve the overall spatial resolution, localization accuracy, and imaging performance of the SIRMIO PET scanner.