数字病理辅助的宫颈癌筛查合理性分析  

Rationality analysis of digital pathology-assisted cervical cancer screening

作　　者：陈莉萍 邓再兴 谢慧君 CHEN Li-Ping;DENG Zai-Xing;XIE Hui-Jun(Department of Pathology,Huzhou Maternity and Child Health Care Hospital,Huzhou,Zhejiang 313000,China)

机构地区：[1]湖州市妇幼保健院病理科,浙江湖州313000

出　　处：《中国妇幼保健》2022年第21期3887-3890,共4页Maternal and Child Health Care of China

基　　金：浙江省医学会临床科研基金项目(2019ZYC-A119)。

摘　　要：目的 探讨数字病理辅助的宫颈癌筛查的合理性。方法 选取2020年6月—2021年5月湖州市妇幼保健院门诊和住院患者宫颈液基细胞学标本,复查最终诊断的300例阴性和200例阳性病例且行高危型人乳头瘤病毒(hr-HPV)mRNA检测的病例,比较人工筛查和数字病理辅助诊断的阳性预测值和阴性预测值。结果 200例阳性病例中,数字病理辅助诊断和人工筛查人员诊断均较好地分类,但由于人工筛查人员对各分级标准的掌握有一定程度的差异,个别病例分类不准确。数字病理辅助诊断200例阳性病例均识别出异常细胞,但对不除外高级别上皮内病变的非典型细胞(ASC-H)、鳞状细胞癌(SCC)及腺上皮病变不能很好地分类,但均能判读为异常细胞,具体分级需要人工复核再分级。数字病理辅助诊断ASC-H、SCC及腺上皮病变的标准尚未形成。结论 人工智能辅助筛查系统具有很好的阳性预测值和阴性预测值等待点,能有效提高工作效率。可进一步积累数据,通过人工智能辅助筛查系统进行深度学习,未来对宫颈癌筛查具有广阔的应用前景。Objective To explore the rationality of digital pathology-assisted cervical cancer screening.Methods Cervical fluid based cell cytology samples of outpatients and inpatients in Huzhou Maternity and Child Health Care Hospital from June 2020 to May 2021 were selected to review the final diagnosis of 300 negative cases and 200 positive cases and high-risk human papillomavirus(hr-HPV)mRNA detection, and to compare the positive predictive value and negative predictive value of manual screening and digital pathology-assisted diagnosis.Results Among the 200 positive cases, digital pathology-assisted diagnosis and manual screening personnel were well classified, but the classification of individual cases was inaccurate due to the differences in the manual screening personnel’s grasp of each classification standard to a certain extent. Abnormal cells were identified in all 200 positive cases by digital pathology-assisted diagnosis, but ASC-H, SCC and glandular epithelial lesions could not be well classified, but all of them could be interpreted as abnormal cells, and the specific classification required manual review and re-classification. The standard of digital pathology-assisted diagnosis of ASC-H, SCC and glandular epithelial lesions has not yet been established.Conclusion The artificial intelligence assisted screening system has good positive predictive value and negative predictive value waiting point, which can effectively improve the work efficiency. It can further accumulate data and conduct deep learning through artificial intelligence-assisted screening system, which has broad application prospects for cervical cancer screening in the future.

关 键 词：宫颈细胞学 宫颈癌筛查 数字病理辅助诊断 人工智能 

分 类 号：R711.74[医药卫生—妇产科学]