Joseph B. Schulz | Searcharxiv

arXiv2026

Non-invasive Reversible Software-based Electron FLASH Irradiation Configuration of a Linear Accelerator in Clinical Use

Configuring clinical linear accelerators (linacs) for ultra-high dose rate (UHDR) electron experiments typically requires invasive hardware manipulation and/or irreversible manufacturer modifications, limiting broader implementation. We present an independently developed UHDR electron configuration of a clinical TrueBeam linac that allows reversible switching between preclinical UHDR and conventional (CONV) modes using only non-invasive software settings. UHDR mode was achieved via service mode software with RF and beam current settings typical of a photon beam, the photon target and monitor chamber retracted, and a clinically unused low-energy scattering foil inserted. An external AC current transformer (ACCT) for beam monitoring, anatomy-specific collimator, and sample holder were mounted on the accessory tray, with external ion chamber in solid water for exit dose monitoring. Percent depth dose (PDD) was measured for UHDR and CONV beams. Dose-per-pulse (DPP) was varied by adjusting gun voltage and quantified with radiochromic film at different source-to-surface distances (SSD). Beam profiles assessed dose uniformity and usable field size. Dose calibration was established between film, ACCT, and ion chamber, and day-to-day reproducibility was tested. PDD confirmed similar energies for UHDR (12.8MeV) and CONV (11.9MeV) beams with matching profiles through mouse thickness. Maximum DPP exceeded 0.5Gy, reaching ~1.5Gy for collimated in vivo setups and ~0.7Gy at extended SSD for tissue culture. Field flatness and symmetry were maintained, supporting organ-specific irradiations and up to 5cm fields for culture. Calibration showed strong linearity across detectors, and output variation was <4%. We demonstrated accurate, reproducible UHDR delivery on a widely available clinical linac with no invasive hardware manipulation, enabling preclinical FLASH research on a clinical treatment machine.

physics.med-ph↗

arXiv2022

Learning Image Representations for Content Based Image Retrieval of Radiotherapy Treatment Plans

Objective: Knowledge based planning (KBP) typically involves training an end-to-end deep learning model to predict dose distributions. However, training end-to-end methods may be associated with practical limitations due to the limited size of medical datasets that are often used. To address these limitations, we propose a content based image retrieval (CBIR) method for retrieving dose distributions of previously planned patients based on anatomical similarity. Approach: Our proposed CBIR method trains a representation model that produces latent space embeddings of a patient's anatomical information. The latent space embeddings of new patients are then compared against those of previous patients in a database for image retrieval of dose distributions. All source code for this project is available on github. Main Results: The retrieval performance of various CBIR methods is evaluated on a dataset consisting of both publicly available plans and clinical plans from our institution. This study compares various encoding methods, ranging from simple autoencoders to more recent Siamese networks like SimSiam, and the best performance was observed for the multitask Siamese network. Significance: Applying CBIR to inform subsequent treatment planning potentially addresses many limitations associated with end-to-end KBP. Our current results demonstrate that excellent image retrieval performance can be obtained through slight changes to previously developed Siamese networks. We hope to integrate CBIR into automated planning workflow in future works, potentially through methods like the MetaPlanner framework.

cs.CV↗