Overview

Precision in radiation therapy is crucial for effectively targeting tumors while minimizing damage to healthy tissue. A new AI tool developed by Northwestern Medicine, named iSeg, has shown the ability to match oncologists in accurately mapping lung tumors on CT scans and even identify areas that may be overlooked by doctors.

Key Findings

• The AI tool iSeg is the first 3D deep learning model capable of segmenting tumors as they move with each breath, which is essential for radiation treatment planning.
• iSeg was trained using CT scans and tumor outlines from hundreds of lung cancer patients across multiple clinics, providing a robust dataset for its development.
• In tests, iSeg consistently matched expert outlines and identified additional high-risk areas that could lead to worse outcomes if untreated.

Expert Insights

Dr. Mohamed Abazeed, senior author and chair of radiation oncology at Northwestern University, stated, “This technology aims to provide doctors with better tools for more precise cancer treatments.” The research team is now working on testing iSeg in clinical settings and plans to expand its application to other tumor types and imaging methods.

Future Directions

The team is also integrating user feedback into iSeg and exploring its adaptation for other imaging techniques, such as MRI and PET scans. Co-author Troy Teo emphasized the potential for this technology to standardize tumor targeting in radiation oncology, especially in areas with limited access to specialized expertise.

Publication Details

The findings will be published in the journal npj Precision Oncology on June 30, 2025.