Researchers created a visual evaluation algorithm that uses deep learning to detect cervical precancer and cancer more accurately than human clinicians, according to a study published in the Journal of the National Cancer Institute.
To train the algorithm, researchers used archived photos of cervixes from a previous study that followed 9,406 women ages 18-94 in Costa Rica for seven years.
The algorithm detected more suspicious areas and cancers than human clinicians visually inspecting the cervix, researchers found. The algorithm also identified more cancers than Pap smear tests.
Researchers said the technology could improve cervical cancer screening and treatment for women in low- and middle-income countries, where 90 percent of cervical cancer deaths occur.