Google developers said this week that the company has trained algorithms to analyze CT scans to find early signs of lung cancer.
At the Google I/O developer conference, Google said its AI model could identify early signs of lung cancer in patients who later developed the disease despite five out of six radiologists missing the signs of the disease. The study will soon be published in the journal Nature Medicine.
“Very early stage cancer is minuscule and can be hard to see, even for seasoned radiologists, which means that many patients with late-stage lung cancer have subtle signs on earlier scans,” Lily Peng, a product manager at Google Brain AI Research, said at the conference. “By looking at many [CT scans], the model learns to detect malignancy, with a performance that meets or exceeds that of trained radiologists.”
Lung cancer is the second most common cancer in adults in the U.S., according to the American Cancer Society. More than 430,000 people who are alive today have been diagnosed with lung cancer. The American Cancer Society predicts there will be an estimated 142,670 deaths from lung cancer this year.
“We know that when cases are diagnosed early, patients have a higher chance of survival. But unfortunately, over 80% of lung cancers are not caught early,” Peng said.
Google has recently been exploring the use of AI in diagnostics and has used the technology to detect metastatic breast cancer and improve grading in prostate cancer cases. Google developers have also trained their AI algorithms to scan eyes for diabetic retinopathy, one of the most common causes of vision loss for people with diabetes.