宁夏回族自治区肺结核人工智能影像筛查系统的应用

Application of Artificial Intelligence Image Screening System for Tuberculosis in Ningxia Autonomous Region

目的:通过肺结核人工智能影像筛查系统解决宁夏回族自治区乡镇卫生院阅读胸片能力不足问题,并探索人工智能影像筛查在肺结核筛检中的价值。方法:采用肺结核人工智能影像筛查系统,在宁夏构建了以省结核病防治所作为中心覆盖各市县乡镇基层卫生院的结核病AI管理系统,并在试点地区评估了该系统的敏感度、特异度及准确度。结果:试点地区肺结核人工智能影像筛查系统诊断肺结核的敏感度、特异度和准确度分别为93.5%(58/62)、86.0%(394/458)、86.9%(452/520)。结论:肺结核智能管理系统很好地解决乡镇卫生院阅读胸片能力不足问题,同时大大提高转诊质量,减少医院不必要的人力和物资的支出及患者交通往返,提升全区结核病防控能力。

Objectives: To solve the insufficient ability of chest films interpretation in Ningxia Autonomous Region Township Hospital by the artificial intelligence image screening system of tuberculosis, and to explore the value of artificial intelligence image screening in tuberculosis screening. Methods:The research uses the tuberculosis artificial intelligence image screening system and constructed a TB AI management system covering all Ningxia township hospitals with the Ningxia provincial TB prevention and control center as the center, and the sensitivity, specificity and accuracy of the system were assessed in the pilot area. Results:The sensitivity, specificity and accuracy of the artificial intelligence of tuberculosis in Ningxia in the pilot area were 93.5%(58/62)、86.0%(394/458) and 86.9%(452/520). Conclusion: The AI management system of tuberculosis can solve the problem of insufficient ability of reading chest film in township hospitals, improve the quality of referral greatly, reduce the unnecessary expenditure of manpower and materials and the transportation of patients and greatly improve TB prevention and control capacity in the whole province.