New AI tool outperforms humans at lung tumor segmentation
Experts have created a tool capable of significantly improving tumor segmentation prior to radiation therapy.
Developed by researchers at Northwestern Medicine in Chicago, the software—known as iSeg—is just as accurate as doctors at outlining tumor margins. In some cases, iSeg even detects areas that doctors have previously overlooked. Experts involved in its development are hopeful its use could lead to improved radiation therapy planning.
“We’re one step closer to cancer treatments that are even more precise than any of us imagined just a decade ago,” said senior author Mohamed Abazeed, MD, PhD, chair and professor of radiation oncology at Northwestern University Feinberg School of Medicine.
To develop iSeg, the team utilized a multi-center registry with imaging data from nine institutions. This provided them with hundreds of CT scans from patients who had been diagnosed with lung cancer, in addition to the manual segmentations of the tumors. Following its training, iSeg was tested on a group of CT scans it had not seen before and tasked with outlining tumor margins.
When compared to doctors’ manual segmentations, iSeg’s performance was consistently on par with that of the professionals; it matched human inter-observer variability and achieved more precise delineation in smaller tumors compared to the physicians, spotting areas that were often missed by the original readers. What’s more, the areas missed by humans were found to be associated with worse outcomes if they went undetected long enough to progress, the group noted.
“Accurate tumor targeting is the foundation of safe and effective radiation therapy, where even small errors in targeting can impact tumor control or cause unnecessary toxicity,” Abazeed suggested.
The next step for iSeg is testing in real-world settings. The team is adding additional features, such as user feedback, and is working to train the model to detect tumors on imaging from different modalities, like MRI and PET, as well. If all goes well and the tool performs as expected, researchers are hopeful it could be deployed in clinical settings “within a couple years.”
Read more about iSeg here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.