A newly developed deep learning algorithm more accurately detected cervical precancer than highly experienced physicians and current testing methods, reported authors of a Jan. 10 study published in the Journal of the National Cancer Institute.
Led by researchers at the National Institutes of Health (NIH), the team used more than 60,000 cervical images taken from 9,406 women during a seven-year cervical cancer screening study performed in Costa Rica. Overall, the algorithm was more accurate at identifying precancerous changes that require medical attention compared to human interpretations and traditional cytology.
"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," said senior author Mark Schiffman, MD, of the National Cancer Institute, in a news release." In fact, the computer analysis of the images was better at identifying precancer than a human expert reviewer of Pap tests under the microscope (cytology)."
The automated system achieved an area under the curve (AUC) score of 0.91, more accurate than the 0.69 AUC achieved by human readers and Pap testing (0.71). The results were independently confirmed by experts at the National Library of Medicine, according to the study.
Importantly, nearly 80 percent of the more than 500,000 annual cervical cancer cases occur in low-resource countries where screening infrastructure is lacking. This method holds promise for bringing low-cost screening to the areas that need it most, the authors argued.
"When this algorithm is combined with advances in HPV vaccination, emerging HPV detection technologies, and improvements in treatment, it is conceivable that cervical cancer could be brought under control, even in low-resource settings," said Maurizio Vecchione, executive vice president of Global Good, a project of intellectual ventures, which helped lead the study.