Arkajyoti Roy, assistant professor in the Department of Management Science and Statistics in the UTSA Carlos Alvarez College of Business is leading a first-of-its kind study working with colleagues at UT Health San Antonio to explore how artificial intelligence (AI) can be used to track cancer tumor shrinkage in patients who receive radiotherapy.
Many studies have shown there is significant tumor shrinkage during radiotherapy. When patients receive radiation treatments, however, doctors typically ignore tumor volume changes, and they deliver the same radiation dose repeatedly during a five-to-seven-week treatment.
The team’s study, titled “Managing tumor changes during radiotherapy using a deep learning model,” has been published in the latest edition of the Medical Physics journal.
The team’s research revealed a better option. It demonstrated that it is possible to accurately predict tumor shrinkage and include those tumor changes in radiation treatment plans. This approach can reduce toxicity to surrounding organs and healthy tissue and ultimately improving the patient’s quality of life.
“With improvements in imaging technologies, we are able to gather higher resolution patient data during cancer radiation treatments. Combined with novel AI algorithms, we developed predictive and prescriptive models that can predict a patient’s anatomy in the future, and more importantly, design adaptive radiotherapy plans in anticipation of the changes,” said Roy.
A significant concern for experts is tracking the tumor shrinkage in lungs during radiotherapy. Using the weekly scans of 16 lung cases, the researchers used AI algorithms to identify the changes in the characteristics in the patient’s anatomy that allowed for the prediction of future tumor shape and size on new patients.
With these AI predictions in hand, the researchers optimized future treatment plans by maximizing the radiation dose coverage in the predicted tumors while minimizing radiation exposure to surrounding organs and healthy tissue.
The results showed that using the deep learning model allowed doctors to significantly reduce the therapy dose and still maintain tumor coverage—a great improvement to current clinical practice.
The research team is now developing an improved AI-based adaptive model to better predict changes in the tumor shape and size to offer medical providers with valuable historical data that will further benefit treatment protocols for patients.
“Our goal is to merge both worlds, the AI predictions and optimization in one model to improve the radiation treatments implemented in the clinic,” Roy concluded.