Retrospective analysis demonstrates improved specificity for detecting adenocarcinoma in Pap tests, highlighting the potential of AI to enhance cervical cancer screening.
Key Points
To assess the diagnostic efficacy and accuracy of detecting atypical glandular cells (AGC) on Pap tests using the Hologic Genius Digital Diagnostics System (HGDDS).
Retrospective analysis of 451 ThinPrep Pap cases (207 AGC, 27 HSIL, 25 LSIL, 192 benign), including 66 histology-confirmed adenocarcinoma cases.
Slides were scanned and analyzed by HGDDS, screened by an experienced cytologist, and independently reviewed by two cytopathologists with access to patient age and HPV status.
Inter-observer concordance between the two cytopathologists showed strong diagnostic agreement (κ = 0.829).
Sensitivity for detecting adenocarcinoma on HGDDS was 98.5% and 95.5% for the two reviewers, comparable to original ThinPrep interpretation (OTPI).
Specificity for adenocarcinoma detection improved significantly with HGDDS (84.6% and 85.6%) compared to OTPI (27.7%), along with higher positive predictive value.