{"title":"人工智能骨髓细胞识别分析系统在血液病辅助诊断中的应用","authors":"Yan Huang, Yun-Ke Wan, Jian-Lan Li","doi":"10.19746/j.cnki.issn.1009-2137.2025.04.041","DOIUrl":null,"url":null,"abstract":"<p><strong>Objective: </strong>To investigate the clinical application value of artificial intelligence (AI)-based bone marrow cell recognition and analysis system in the diagnosis of hematological diseases.</p><p><strong>Methods: </strong>The bone marrow smears of hematological patients who were admitted to The Second Hospital of Shanxi Medical University from 2018 to 2020 were retrospectively analyzed. A total of 115 bone marrow smears with clear diagnosis and typical cell morphology characteristics were selected, including 20 cases of immune thrombocytopenia(ITP), 11 cases of iron deficiency anemia (IDA), 17 cases of megaloblastic anemia (MA), 20 cases of chronic myeloid leukemia (CML), 17 cases of acute lymphoblastic leukemia (ALL), 23 cases of acute promyelocytic leukemia (APL), and 7 cases of acute myeloid leukemia unclassified (AML-M2). The samples were analyzed by manual microscopic examination, AI automatic recognition, and manual correction after AI recognition.</p><p><strong>Results: </strong>The images captured by the AI device were clear, and the cell morphological structures were distinct. The average experimental diagnostic efficiency parameters of the bone marrow nucleated cells classified in this system were calculated. The sensitivity was 74.90%, specificity was 99.03%, and accuracy was 98.29%. In the comparison between the AI recognition group and the manual examination group, the data of IDA, ITP, MA, and CML diseases were all greater than 0.85 in ICC correlation coefficient, with excellent consistency; the data of APL, AML-M2, and ALL three diseases were between 0.6 and 0.85 in ICC correlation coefficient, with moderate consistency. However, after manual review and correction, the ICC correlation coefficient between the data of the AI correction group and the data from the manual examination group was greatly improved.</p><p><strong>Conclusion: </strong>The AI bone marrow cell recognition and analysis system has the characteristics of high accuracy, high specificity, good sensitivity and fast detection. When used in combination with manual review, it can improve the detection efficiency of bone marrow cells morphological analysis and meet the needs of clinical work.</p>","PeriodicalId":35777,"journal":{"name":"中国实验血液学杂志","volume":"33 4","pages":"1203-1208"},"PeriodicalIF":0.0000,"publicationDate":"2025-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"[Application of Artificial Intelligence Bone Marrow Cell Recognition and Analysis System in Auxiliary Diagnosis of Hematological Disease].\",\"authors\":\"Yan Huang, Yun-Ke Wan, Jian-Lan Li\",\"doi\":\"10.19746/j.cnki.issn.1009-2137.2025.04.041\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Objective: </strong>To investigate the clinical application value of artificial intelligence (AI)-based bone marrow cell recognition and analysis system in the diagnosis of hematological diseases.</p><p><strong>Methods: </strong>The bone marrow smears of hematological patients who were admitted to The Second Hospital of Shanxi Medical University from 2018 to 2020 were retrospectively analyzed. A total of 115 bone marrow smears with clear diagnosis and typical cell morphology characteristics were selected, including 20 cases of immune thrombocytopenia(ITP), 11 cases of iron deficiency anemia (IDA), 17 cases of megaloblastic anemia (MA), 20 cases of chronic myeloid leukemia (CML), 17 cases of acute lymphoblastic leukemia (ALL), 23 cases of acute promyelocytic leukemia (APL), and 7 cases of acute myeloid leukemia unclassified (AML-M2). The samples were analyzed by manual microscopic examination, AI automatic recognition, and manual correction after AI recognition.</p><p><strong>Results: </strong>The images captured by the AI device were clear, and the cell morphological structures were distinct. The average experimental diagnostic efficiency parameters of the bone marrow nucleated cells classified in this system were calculated. The sensitivity was 74.90%, specificity was 99.03%, and accuracy was 98.29%. In the comparison between the AI recognition group and the manual examination group, the data of IDA, ITP, MA, and CML diseases were all greater than 0.85 in ICC correlation coefficient, with excellent consistency; the data of APL, AML-M2, and ALL three diseases were between 0.6 and 0.85 in ICC correlation coefficient, with moderate consistency. However, after manual review and correction, the ICC correlation coefficient between the data of the AI correction group and the data from the manual examination group was greatly improved.</p><p><strong>Conclusion: </strong>The AI bone marrow cell recognition and analysis system has the characteristics of high accuracy, high specificity, good sensitivity and fast detection. When used in combination with manual review, it can improve the detection efficiency of bone marrow cells morphological analysis and meet the needs of clinical work.</p>\",\"PeriodicalId\":35777,\"journal\":{\"name\":\"中国实验血液学杂志\",\"volume\":\"33 4\",\"pages\":\"1203-1208\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2025-08-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"中国实验血液学杂志\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.19746/j.cnki.issn.1009-2137.2025.04.041\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"Medicine\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"中国实验血液学杂志","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.19746/j.cnki.issn.1009-2137.2025.04.041","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"Medicine","Score":null,"Total":0}
[Application of Artificial Intelligence Bone Marrow Cell Recognition and Analysis System in Auxiliary Diagnosis of Hematological Disease].
Objective: To investigate the clinical application value of artificial intelligence (AI)-based bone marrow cell recognition and analysis system in the diagnosis of hematological diseases.
Methods: The bone marrow smears of hematological patients who were admitted to The Second Hospital of Shanxi Medical University from 2018 to 2020 were retrospectively analyzed. A total of 115 bone marrow smears with clear diagnosis and typical cell morphology characteristics were selected, including 20 cases of immune thrombocytopenia(ITP), 11 cases of iron deficiency anemia (IDA), 17 cases of megaloblastic anemia (MA), 20 cases of chronic myeloid leukemia (CML), 17 cases of acute lymphoblastic leukemia (ALL), 23 cases of acute promyelocytic leukemia (APL), and 7 cases of acute myeloid leukemia unclassified (AML-M2). The samples were analyzed by manual microscopic examination, AI automatic recognition, and manual correction after AI recognition.
Results: The images captured by the AI device were clear, and the cell morphological structures were distinct. The average experimental diagnostic efficiency parameters of the bone marrow nucleated cells classified in this system were calculated. The sensitivity was 74.90%, specificity was 99.03%, and accuracy was 98.29%. In the comparison between the AI recognition group and the manual examination group, the data of IDA, ITP, MA, and CML diseases were all greater than 0.85 in ICC correlation coefficient, with excellent consistency; the data of APL, AML-M2, and ALL three diseases were between 0.6 and 0.85 in ICC correlation coefficient, with moderate consistency. However, after manual review and correction, the ICC correlation coefficient between the data of the AI correction group and the data from the manual examination group was greatly improved.
Conclusion: The AI bone marrow cell recognition and analysis system has the characteristics of high accuracy, high specificity, good sensitivity and fast detection. When used in combination with manual review, it can improve the detection efficiency of bone marrow cells morphological analysis and meet the needs of clinical work.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
