In the realm of radiation therapy, precision is paramount in saving lives. Oncologists rely on accurately mapping tumor size and location before administering high-dose radiation to eradicate cancer cells while preserving healthy tissue. Traditionally, this process, known as tumor segmentation, has been performed manually, leading to variations between doctors and the potential oversight of critical tumor areas. However, a groundbreaking AI tool named iSeg, developed by a team of Northwestern Medicine scientists, is changing the game. Not only does iSeg match expert oncologists in outlining lung tumors on CT scans, but it also identifies areas that may be overlooked by some doctors, as revealed in a recent comprehensive study. Unlike previous AI tools that focused on static images, iSeg is the first 3D deep learning tool capable of segmenting tumors in real-time as they move with each breath. This feature is crucial in planning radiation treatment, a procedure that half of all cancer patients in the U.S. undergo during their illness. The implications of this cutting-edge technology are profound, bringing us closer to cancer treatments that are more precise than ever imagined just a decade ago. Dr.
Mohamed Abazeed, the senior author of the study and a prominent figure in radiation oncology at Northwestern University Feinberg School of Medicine, emphasizes the goal of providing doctors with superior tools for personalized and improved cancer treatment. The AI was trained using CT scans and tumor outlines drawn by doctors from hundreds of lung cancer patients across multiple clinics within the Northwestern Medicine and Cleveland Clinic health systems. This extensive dataset surpasses the limited, single-hospital data used in previous studies. Following training, iSeg was tested on unseen patient scans, where its tumor outlines were compared to those delineated by physicians. The study demonstrated that iSeg consistently aligned with expert outlines from various hospitals and scan types. Furthermore, it detected additional areas that some doctors had missed, which were associated with adverse outcomes if left untreated. This underscores the potential of iSeg in identifying high-risk regions that often evade detection. ‘Accurate tumor targeting is fundamental to safe and effective radiation therapy, as even minor errors in targeting can impact tumor control or lead to unnecessary toxicity,’ remarks Dr. Abazeed. The advancement of AI technology in tumor segmentation not only streamlines the treatment process but also enhances the precision and efficacy of cancer therapies, marking a significant stride towards personalized and optimized patient care.
Read more from sciencedaily.com