With support from the National Institutes of Health, the Lin Lab works to accelerate SIEMAC (Simultaneous Intensity Energy Modulation and Compensation) using graphics processing units for community use, and applies deep learning and Monte Carlo methods to validate particle therapy microdosimetry.
This figure illustrates the SIEMAC (Simultaneous Intensity Energy Modulation and Compensation) framework and representative results, demonstrating a novel proton FLASH optimization strategy. The method builds on Harrison et al. (IJROBP 2023) and forms the basis of Specific Aim 1 of the awarded NIH grant.
This figure presents the HPT (Hybrid Particle Track) U-net architecture designed to separate overlapping particle tracks that accumulate intensities, a task distinct from conventional anatomic segmentation. Shown are example inputs, network outputs, and validation from the PTCOG 2026 oral presentation; this work supports Specific Aim 2 of the awarded NIH grant and extends prior work published in Medical Physics (2024).
This figure summarizes the STEAL (Simultaneous Track Energy and Length) extension to Monte Carlo simulations, enabling inference of measured particle species by jointly modeling track length and deposited energy. The approach differentiates primary protons, secondary protons, and beta particles based on track morphology; the manuscript describing this method is currently in preparation.
This figure shows preliminary STEAL results applied to data from a pixelated semiconductor detector developed using CERN technology. Linear energy transfer spectra are shown for transmitted primary protons and for secondary particles (light ions, secondary protons, and beta particles) that fully stop within the detector. (Detector details)