Grant Funding (03/2026): Investigator-Initiated Varian Support for AI Digital Twins
New Investigator-Initiated Funding
Varian (Siemens Healthineers), March 2026
The Thomas Lab has received investigator-initiated research funding from Varian (Siemens Healthineers) to support a new project developing AI-driven digital twins of patients during radiation therapy.
This takes total lab funding to over $1M. The project goal is to use camera-based computer vision and modern AI models to build a real-time digital representation of patient anatomy during treatment. Digital twins can continuously track body position and motion to help ensure radiation is delivered safely and precisely.
We also expect this core technology to have broader applications in medicine and education, including clinical training and simulation.
Abstract: Depth from Vision: Computer Vision for Anatomy-Aware SGRT
Our project evaluates the first TJU in-house-developed artificial intelligence (AI) algorithm designed to reduce treatment-related side effects in cancer patients. Surface-guided radiotherapy (SGRT) is used in breast cancer treatment to reduce radiation exposure to the heart and lungs by tracking patient position and breathing during treatment.
However, SGRT is inconsistently used due to high equipment costs, limited robustness, and reduced positioning confidence. Approximately 40% of otherwise eligible patients cannot currently be treated using SGRT systems. We have developed a novel AI and computer vision method that combines standard cameras with a neural network to guide patient positioning during radiotherapy.
This low-cost approach aims to improve positioning accuracy and broaden SGRT applicability to a more diverse patient population. The proposed system uses multi-camera video acquired during routine CT simulation and treatment to reconstruct patient motion and guide positioning during radiotherapy.
These data will be used to evaluate agreement between the proposed approach and existing clinical SGRT systems in measuring patient motion. Results from this work will support expansion of reliable SGRT use across the Jefferson enterprise and inform future multi-site clinical studies.