Advancing Hyperspectral Targeted Alpha Therapy with Adversarial Machine Learning

Abstract

Targeted Alpha Therapy (TAT) has emerged as a promising modality for the treatment of various malignancies, leveraging the high linear energy transfer (LET) and short range of alpha particles to selectively irradiate cancer cells while sparing healthy tissue. Monitoring and optimizing TAT delivery is crucial for its clinical success. Hyper-spectral Single Photon Imaging (HSPI) presents a novel and versatile approach for the real-time assessment of TAT in vivo. This study introduces a comprehensive framework for HSPI in TAT, encompassing spectral unmixing, quantitative dosimetry, and spatiotemporal visualization. We report the development of a dedicated HSPI system tailored to alpha-emitting radionuclides, enabling the simultaneous acquisition of high-resolution spectral data and single-photon localization. Utilizing advanced spectral unmixing algorithms, we demonstrate the discrimination of alpha-induced scintillation from background fluorescence, facilitating precise alpha particle tracking with adversarial machine learning.