The National Institutes of Health and Global Good has developed an AI algorithm which can identify cervical precancer changes which need attention.
The AI approach, called automated visual evaluation, could revolutionise cervical cancer screening by allowing for the analysis of digital images of a woman’s cervix to identify cervical precancer.
Mark Schiffman, M.D., M.P.H., of NCI’s Division of Cancer Epidemiology and Genetics, and senior author of the study, said: “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. 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).”
Cervical precancer identification in low-resource settings
The new method is especially helpful in low-resource settings. Healthcare workers in these settings currently use visual inspection with acetic acid (VIA), a screening method.
During this method, a health worker applies dilute acetic acid to the cervix and inspects the cervix with their naked eye to look for “aceto whitening” which indicates possible disease. VIA is widely used where more advanced screening methods are not available due to convenience and low cost. However, it is known to be inaccurate and needs improvement.
The AI algorithm
The AI approach is similarly easy to perform. Health workers can use a mobile phone or similar camera device for cervical screening and treatment during a visit. This can be performed with minimal training, making it ideal for countries with limited health care resources, where cervical cancer is a leading cause of illness and death in women.
The study found that the AI algorithm performed better than all standard screening tests at predicting all cases diagnosed. The AI identified cervical precancer with greater accuracy (AUC=0.91) than a human expert review (AUC=0.69) or conventional cytology (AUC=0.71).
According to the authors, an AUC of 0.5 indicates a test that is no better than chance, whereas an AUC of 1.0 represents a test with perfect accuracy in identifying disease.