Cervical cytology screening facilitated by an artificial intelligence microscope: A preliminary study.

摘要: BACKGROUND Cervical cytology screening is usually laborious with a heavy workload and poor diagnostic consistency. The authors have developed an artificial intelligence (AI) microscope that can provide onsite assistance for cervical in real time. METHODS A total of 2167 slides were selected from cohort 10,601 cases Shenzhen Maternity Child Healthcare Hospital, the training data set consisted 42,073 abnormal epithelial cells. recognition results AI technique presented eyepiece by augmented reality technique. Potentially cells highlighted binary classification 10× field view (FOV) multiclassification according to Bethesda system 20× 40× FOVs. In addition, 486 reader study evaluate performance microscope. RESULTS study, which compared manual reading assistance, sensitivities detection low-grade squamous intraepithelial lesions high-grade significantly improved 0.837 0.923 (P < .001) 0.830 0.917 .01), respectively; κ score atypical undetermined significance (ASCUS) was 0.581 0.637; averaged pairwise consistency 0.649 0.706; 0.720 0.798; ASCUS 0.557 0.639. CONCLUSIONS this show real-time improve efficiency accuracy diagnosis.