人工智能辅助血细胞形态学检查的技术要求及其临床应用中国专家共识(2024年版)  

Chinese expert consensus on the technical and clinical practice specifications of artificial intelligence assisted morphology examination of blood cells (2024)

作　　者：中华医学会血液学分会实验诊断学组 吴德沛[2] 张曦 陈苏宁[2] 彭贤贵 刘水清 刘丹丹[5] 杨武晨 王翊[6] Laboratory Diagnosis Group,Chinese Society of Hematology,Chinese Medical Association;Wu Depei;Zhang Xi;Chen Suning;Peng Xiangui(不详;the First Affiliated Hospital of Soochow University,Jiangsu Institute of Hematology,National Clinical Research Center for Hematologic Diseases,Suzhou 215006,China;the Second Affiliated Hospital of Army Medical University,Medical Center of Hematology,Chongqing 400037,China;the First Afiliated Hospital of Soochow University,Jiangsu Institute of Hematology,National Clinical Research Center for Hematologic Diseases,Suzhou 215006,China;the Second Afiliated Hospital of Army Medical University,Medical Center of Hematology,Chongqing 400037,China)

机构地区：[1]不详 [2]苏州大学附属第一医院,江苏省血液研究所,国家血液系统疾病临床医学研究中心,苏州215006 [3]陆军军医大学第二附属医院,血液病医学中心,重庆400037 [4]陆军军医大学第二附属医院 [5]苏州大学附属第一医院 [6]重庆大学计算机学院

出　　处：《中华血液学杂志》2024年第4期330-338,共9页Chinese Journal of Hematology

摘　　要：血细胞形态学检查是血液疾病诊断的重要手段,传统人工镜检存在效率低、易受主观影响等问题。人工智能技术的应用提高了血细胞检查的效率和质量,促进了检测结果的标准化。目前,多种人工智能设备已在临床使用或研究中,但其技术要求和配置各异。中华医学会血液学分会实验诊断学组组织相关专家,制订了本共识。本共识内容包括术语定义、适用范围、技术要求、临床应用、数据管理和信息安全,强调了标本制备、图像采集、图像分割算法、细胞特征提取与识别分类的重要性,并提出了细胞识别谱的基本要求。同时,对病理细胞的细分类、细胞训练与测试的要求、质量控制标准、辅助人工出具诊断报告等方面进行了详细说明。此外,还强调了数据管理和信息安全的重要性,确保患者信息的安全和数据的准确性。Blood cell morphological examination is a crucial method for the diagnosis of blood diseases,but traditional manual microscopy is characterized by low efficiency and susceptibility to subjective biases.The application of artificial intelligence(AI)technology has improved the efficiency and quality of blood cell examinations and facilitated the standardization of test results.Currently,a variety of AI devices are either in clinical use or under research,with diverse technical requirements and configurations.The Experimental Diagnostic Study Group of the Hematology Branch of the Chinese Medical Association has organized a panel of experts to formulate this consensus.The consensus covers term definitions,scope of application,technical requirements,clinical application,data management,and information security.It emphasizes the importance of specimen preparation,image acquisition,image segmentation algorithms,and cell feature extraction and classification,and sets forth basic requirements for the cell recognition spectrum.Moreover,it provides detailed explanations regarding the fine classification of pathological cells,requirements for cell training and testing,quality control standards,and assistance in issuing diagnostic reports by humans.Additionally,the consensus underscores the significance of data management and information security to ensure the safety of patient information and the accuracy of data.

关 键 词：血细胞形态学检查 人工智能 临床应用 数据管理与信息安全 细胞识别谱 

分 类 号：R55[医药卫生—血液循环系统疾病]