AI tool could help predict side effects from lung cancer treatment

Radiation therapy is a cornerstone of lung cancer treatment. But even when delivered with precision, radiation can damage healthy lung tissue.

“Try as we might, when we deliver radiation to a cancer, some goes to surrounding organs and patients can have side effects including lung inflammation, coughing and feeling short of breath,” says medical physicist Yevgeniy Vinogradskiy, PhD.

Now, Dr. Vinogradskiy and a team of students, clinicians and computer scientists have developed a tool to help identify who might be at risk for those side effects before radiation treatment begins.

The study, published in Reports of Practical Oncology and Radiotherapy, focuses on the lung’s natural subunits, called lobes. Rather than treating the lung as a single organ, researchers have found that the radiation dose administered to specific lobes matters. Damage to certain lobes, particularly the lower ones, raises the risk of lung toxicity and side effects.

To know how much radiation each lobe receives, clinicians draw the outline of each lobe by hand on CT scans. The process can take three to four hours per patient. As a result, lobe-level analysis is often skipped, even though it could provide more accurate predictions of side effects.

To solve this problem, Dr. Vinogradskiy and colleagues trained a machine learning model to automatically identify and outline all five lung lobes on a scan. The team trained the AI using scans from 40 lung cancer patients from two institutions where specialists had already identified each lobe. Then the researchers tested the model on 10 scans it had never seen before. The results were promising.

“A machine learning model can identify lung lobes just as accurately as a clinician,” says Dr. Vinogradskiy. “But instead of hours, it takes less than a second.”

By making lobe-level analysis practical, the tool opens the door to better predicting side effects from radiation in lung cancer patients.

“Knowing the future is powerful,” says Dr. Vinogradskiy, a member of Sidney Kimmel Medical College. “If we can identify which patients are at higher risk, we can intervene earlier and potentially prevent serious complications.”

By Roni Dengler