THURSDAY, Jan. 10, 2019 (HealthDay News) -- A deep learning-based visual evaluation algorithm can detect cervical precancer/cancer with higher accuracy than conventional cytology, according to a study published online Jan. 10 in the Journal of the National Cancer Institute.
Liming Hu, Ph.D., from the Intellectual Ventures Global Good Fund in Bellevue, Washington, and colleagues followed a population-based longitudinal cohort of 9,406 women ages 18 to 94 years from 1993 to 2000. They incorporated multiple cervical screening methods and histopathologic confirmation of precancers. Cancers were identified for up to 18 years of extended follow-up using tumor registry linkage. The deep learning-based algorithm was trained/validated using archived, digitized cervical images from screening (cervigrams). For detection of precancer/cancer, the resultant image prediction score (0 to 1) could be categorized to balance sensitivity and specificity.
The researchers found that cumulative precancer/cancer cases were identified with greater accuracy using automated visual evaluation of enrollment cervigrams compared with the original cervigram interpretation or conventional cytology (area under the curve, 0.91 versus 0.69 and 0.71, respectively). A total of 55.7 percent of 228 precancers diagnosed cumulatively in the entire adult population were identified using a single visual screening round restricted to women at the prime screening ages of 25 to 49 years; 11.0 percent were referred for management.
"Our findings show that a deep learning algorithm can use images collected during routine cervical cancer screening to identify precancerous changes that, if left untreated, may develop into cancer," a coauthor said in a statement.
Abstract/Full Text
Editorial
