A team of Google AI-led researchers have created an algorithm that predicts a patients risk of lung cancer from CT scans, according to a May 20 study published in Nature Medicine.
The deep learning platform, tested on more than 6,000 cases from the National Lung Cancer Screening Trial and Northwestern University, performed similarly to six radiologists.
“Despite the value of lung cancer screenings, only 2-4 percent of eligible patients in the U.S. are screened today,” co-author Shravya Shetty, MS, with Google AI, Mountain View, California, wrote in a Google blog post. “This work demonstrates the potential for AI to increase both accuracy and consistency, which could help accelerate adoption of lung cancer screening worldwide.”
The team leveraged 45,856 de-identified chest CT screening exams from the National Lung Cancer Screening Trial. The algorithm was then tested on 6,716 CTs and finally validated on an independent clinical set of 1,139 cases.
Overall, the model achieved a 94.4% area under the curve, similar to the six radiologists. When prior CT imaging results weren’t available, the algorithm beat out all radiologists.
Additionally, Shetty et al. noted, the algorithm reduced false positives by 11% and false negatives by 5%.
“This creates an opportunity to optimize the screening process via computer assistance and automation,” the authors wrote. “While the vast majority of patients remain unscreened, we show the potential for deep learning models to increase the accuracy, consistency and adoption of lung cancer screening worldwide.”
“These initial results are encouraging, but further studies will assess the impact and utility in clinical practice," Shetty added.