Objective: To investigate the efficacy of the PAPNET Testing System and its ability to detect significant areas on clinically important false negative gynecologic smears.
Study design: Sixty-two gynecologic smears that had been obtained from women studied in a previous case-control investigation, completed in 1987, and had originally been interpreted as negative were rescreened by two independent, blind cytotechnologist-cytopathologist teams. Twenty-nine of these "negative" smears were from 19 women who had been subsequently diagnosed with invasive squamous cell carcinoma and had self-reported a history of only negative gynecologic smears. Thirty-three smears were from 33 control women who did not develop cervical cancer. One team, at the University of Southern California (USC), manually rescreened the smears as part of the original study. The other team, at the University of California at Los Angeles (UCLA), recently used the PAPNET Testing System to rescreen the same smears. This computer-assisted system utilizes neural network technology to recognize and select potentially abnormal cell scenes on a conventionally prepared gynecologic smear. The PAPNET-selected scenes are displayed for review by trained cytologists, who ultimately diagnose the smear.
Results: Manual reevaluation of the smears by the USC team in 1987 resulted in the reclassification of 9 of the 29 case smears (31%) and 2 of 33 control smears (6%) as class II to V (atypical squamous cells of undetermined significance to invasive carcinoma). Using the PAPNET System to scan and review the same smears, the cytotechnologist at UCLA referred 24 case smears to the cytopathologist, who ultimately reclassified 12 of the 29 case smears (41%) and 5 of the 33 control smears (15%) as abnormal.
Conclusion: This study supports the use of the PAPNET System as an effective, routine rescreener for the detection of clinically significant false negative gynecologic smears.