Doctors can use artificial intelligence to determine your chance of lung cancer with a single CT scan.
"By catching lung cancer early, we increase the overall survival," said Dr. Florian Fintelmann, a radiologist at Massachusetts General Hospital and associate professor of radiology at Harvard Medical School.
Researchers recently created Sybil, an AI model that can predict someone's future lung cancer risk from a single CT scan.
A study on the model and the algorithm itself was developed by the Mass General Cancer Center and MIT.
"What it does, it enables us to put people who undergo lung cancer screening in different buckets: High risk, medium risk, low risk," Dr. Fintelmann, a co-author of the study, said.
Lung cancer is the second most common cancer in the U.S., according to the American Cancer Society.
The society estimates about 127,070 deaths from lung cancer will occur in the U.S. in 2023.
The study on Sybil found the model can detect lung cancer risk with good accuracy, and predict things doctors can't see.
Sybil was validated using three different data sets, which consisted of more than 20,000 CT scans.
The algorithm doesn't need any additional information like risk factor, age, or gender.
"Sybil can make accurate predictions based on a single time point. So other tools, they need at least a couple of CT scans to compare to each other looking for differences between them," Dr. Fintelmann said.
"One of the questions I sometimes get is 'Will this replace the physician?' And it will definitely not," he said.
While AI can predict your future risk, doctors will still be needed to determine where the cancer is, what it looks like, and possible treatment plans.
"The algorithm is freely available," he said.
However, it may be a while before this tool is used to analyze your next CT scan.
"Right now, the Sybil algorithm is not implemented in the hospital because there's no FDA clearance," Dr. Fintelmann said.
So what's next?
"There are clinical trials that are on the way that will look at the utility and the benefit to patients of using Sybil in clinical decision making for lung cancer screening," he said.
Dr. Fintelmann said they also want to diversify the pool of CT scans the algorithm is trained on, as he sees potential issues for bias based on the data that has been used so far.
Trending stories at Scrippsnews.com
