Ethan Stolen and Assistant Professor of Radiation Oncology James Sohn just published a new paper in Physics and Imaging in Radiation Oncology (phiRO). This is Ethan’s first paper published in this high-impact journal published by the European Society for Radiotherapy and Oncology (ESTRO).
Ethan presented this research at the 2025 ESTRO Congress in Vienna, Austria. Their abstract was selected among the highest scoring to be presented at the conference, and they were invited to publish their full article in a special edition of phiRO, “Physics Highlights from ESTRO 2025”.
Their paper is the latest in their strong collaboration with Mayo Clinic in Florida and Yonsei University in Seoul, bringing together expertise in proton therapy physics, clinical delivery, and machine learning.
“What excited me most about this research was the chance to work with proton therapy experts at one of the nation’s top centers,” Ethan said. “I was honored to have the opportunity to work on a practical project that could directly impact the safety of patient care.”
The paper tackles a fundamental quality assurance challenge in pencil beam scanning proton therapy: even small deviations between planned and delivered proton spot positions can produce clinically meaningful dose differences, particularly in regions with steep dose gradients. Existing log-file–based QA workflows are largely retrospective and offer limited support for catching delivery deviations before the next treatment fraction. To address this gap, the team developed a machine-learning-based QA framework using XGBoost regression models trained on delivery log files and DICOM treatment plans from a Hitachi PROBEAT-V synchrotron-based proton system, with separate models for the x- and y-coordinates of each proton spot.
"This work is a step toward more predictive, data-driven QA in proton therapy," said Dr. Sohn. "Our new tool can give physicists valuable insight into how the machine is actually behaving and help ensure that what we plan is what the patient receives.”
Read the paper here: Yoo, S. K., Yaddanapudi, S., Lu, B., Stolen, E., Sen, S., Choi, B., Kim, J. S., Furutani, K., Beltran, C., & Sohn, J. J. (2026). A feasibility study on a machine-learning-based quality assurance tool for spot-scanning proton therapy using delivery log files and treatment plans. Physics and Imaging in Radiation Oncology, 100987. https://doi.org/10.1016/j.phro.2026.100987