A research team from Northwestern University has developed a new artificial intelligence (AI) tool called iSeg, which has proven its ability to detect lung tumors from CT scans with accuracy comparable to that of specialist doctors, a promising step toward improving the accuracy of cancer treatment.
Unlike previous AI tools that focused on static images, this tool is the first of its kind to rely on 3D deep learning and demonstrates its ability to track tumor movement as a patient breathes, a key element in improving accurate radiation planning.
The tool was trained using hundreds of CT scans and tumor maps drawn by physicians from nine treatment centers within the Northwestern Medicine and Cleveland Clinic systems, allowing the tool a broader range of learning than most previous studies, which relied on data from only one hospital.
Furthermore, the tool was tested on images from patients that had not been seen before and compared with the physicians' results. The study found that the tool accurately matched the tumor boundaries defined by experts, regardless of the hospital or type of scan.
It also revealed additional areas overlooked by some doctors. These areas, if left untreated, lead to worse health outcomes, demonstrating the tool's ability to detect high-risk areas and enhancing its potential to improve survival rates.
Researchers are currently working to test the tool in a real-world clinical setting and expand its use to include other tumor types, as well as adapting it to work with other imaging techniques. This is aimed at standardizing care across institutions and improving patient outcomes.
.jpg)
0 Comments