Researchers at Northwestern University have developed iSeg, a cutting-edge three-dimensional deep learning model that matches radiation oncologists in identifying lung tumors from CT scans. Published in npj Precision Oncology, iSeg automates tumor segmentation, significantly reducing variability in manual mapping. It consistently outlines tumors and identifies high-risk regions often overlooked by clinicians, enhancing treatment outcomes. This model addresses the challenges of tumor movement during breathing, improving accuracy in lung cancer treatment. With radiation therapy being crucial for nearly half of U.S. cancer patients, iSeg’s ability to streamline tumor mapping could enhance care, particularly in settings with limited expertise. Current evaluations are aimed at real-time clinical use, with plans to expand its applications to other organs and imaging methods like MRI and PET. The integration of iSeg into clinical workflows could revolutionize decision support for radiation oncologists in the near future.
For more information, refer to the study by Sarkar et al. (2025).
