人工智能在血液疾病形态学诊断中的应用进展

doi:10.16150/j.1671-2870.2026.01.003

诊断学理论与实践 ›› 2026, Vol. 25 ›› Issue (01): 15-20.doi: 10.16150/j.1671-2870.2026.01.003

人工智能在血液疾病形态学诊断中的应用进展

王琰(), 范磊

江苏省人民医院（南京医科大学附属第一医院）血液内科，南京 210029

收稿日期:2025-12-09 修回日期:2026-01-04 接受日期:2026-01-08 出版日期:2026-02-25 发布日期:2026-02-25
通讯作者: 王琰 E-mail： zx230889zx@163.com
基金资助:
国家自然科学基金面上项目(82470186);江苏省自然科学基金省市联合资助项目(BK20232039)

Application progress of artificial intelligence in morphological diagnosis of blood diseases

WANG Yan(), FAN Lei

Department of Hematology, Jiangsu Province Hospital (The First Affiliated Hospital with Nanjing Medical University), Jiangsu Nanjing 210029, China

Received:2025-12-09 Revised:2026-01-04 Accepted:2026-01-08 Published:2026-02-25 Online:2026-02-25

摘要/Abstract

摘要：

随着全切片扫描的普及与数字病理技术的成熟，采用人工智能（artificial intelligence， AI）技术分析外周血与骨髓涂片已获得了一定的突破，包括目标检测[（You Dnly Look Once， YOLO）等]、弱监督对比学习及多实例学习在解决细胞识别、领域偏移中的应用。其中，多实例学习已被成功应用于急性早幼粒细胞白血病（M₃）的辅助诊断，通过对整张涂片的全局特征聚合，有效实现了对危急血液疾病重症的早期预警。研究表明，AI在骨髓增殖性肿瘤亚型判定中的准确率高达93.1%；AI基于形态特征预测骨髓增生异常综合征（myelodysplastic syndrome， MDS）患者向AML转化风险的受试者操作特征曲线下面积为0.81；AI对造血组织的自动量化结果与病理专家评估结果间的相关性高（r=0.78），证实了AI在骨髓活检病理分析中的价值，尤其是AI在全切片级模型应用于临床的潜力。在临床实践应用方面，目前在外周血涂片分析中，已形成了AI预分类结合人工复核的标准化模式；而在骨髓活检病理分析中，AI虽实现了数字化扫描，但受制于细胞谱系复杂性及病态造血的异质性，目前的应用范围仍局限于细胞计数和初筛。此外，数字化远程会诊在缓解医疗资源分布不均问题中也起了重要作用。未来，随着多模态融合、大语言模型生成报告技术的成熟，AI有望从单纯的计数分类工具进化为集诊断、分型与预后评估于一体的综合性大模型。

关键词: 人工智能, 血液疾病, 形态学, 外周血涂片, 骨髓活检病理

Abstract:

With the widespread adoption of whole-slide scanning and the advancement of digital pathology techno-logy, the use of artificial intelligence (AI) for the analysis of peripheral blood and bone marrow smears has achieved notable breakthroughs, including the application of object detection (You Only Look Once, YOLO), weakly supervised contrastive learning, and multiple instance learning (MIL) in addressing cell recognition and domain shifts. Among them, MIL has been successfully applied in the auxiliary diagnosis of acute promyelocytic leukemia (M₃), effectively providing early warning of critical cases through global feature aggregation of the entire smear. Studies indicate that the accuracy of AI in determining subtypes of myeloproliferative neoplasm (MPN) is as high as 93.1%. The area under the receiver operating characteristic curve for AI predicting the risk of transformation to AML in patients with myelodysplastic syndrome (MDS) based on morphological features is 0.81. The correlation coefficient between AI-based automated quantification of hematopoietic tissue and pathologist assessment results reaches 0.78, confirming the value of AI in bone marrow biopsy pathological analysis, especially highlighting the potential of whole-slide-level models for clinical application. In terms of clinical practice application, a standardized model of AI pre-classification combined with manual review has been established for peripheral blood smear analysis. In bone marrow biopsy pathological analysis, although AI has achieved digital scanning, its current applications remain limited to cell counting and preliminary screening due to the complexity of cell lineages and the heterogeneity of dysplastic hematopoiesis. Furthermore, digital remote consultation plays an important role in alleviating the uneven distribution of medical resources. In the future, with the maturation of multi-modal fusion and large language model (LLM)-based report generation technology, AI is expected to evolve from a simple counting and classification tool into a comprehensive model integrating diagnosis, subtyping, and prognosis assessment.

Key words: Artificial intelligence, Hematopathy, Morphology, Peripheral blood smear, Bone marrow biopsy

中图分类号:

R445
R733.7

王琰, 范磊. 人工智能在血液疾病形态学诊断中的应用进展[J]. 诊断学理论与实践, 2026, 25(01): 15-20.

WANG Yan, FAN Lei. Application progress of artificial intelligence in morphological diagnosis of blood diseases[J]. Journal of Diagnostics Concepts & Practice, 2026, 25(01): 15-20.

参考文献 19

[1]	LU M Y, WILLIAMSON D F K, CHEN T Y, et al. Data-efficient and weakly supervised computational pathology on whole-slide images[J]. Nat Biomed Eng, 2021, 5(6):555-570. doi: 10.1038/s41551-020-00682-w pmid: 33649564
[2]	ECKARDT J N, MIDDEKE J M, RIECHERT S, et al. Deep learning detects acute myeloid leukemia and predicts NPM1 mutation status from bone marrow smears[J]. Leukemia, 2022, 36(1):111-118. doi: 10.1038/s41375-021-01408-w
[3]	CHEN T, KORNBLITH S, NOROUZI M, et al. A simple framework for contrastive learning of visual representations[J/OL]. [2025-12-09]. http://arxiv.org/pdf/2002.05709.
[4]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE,2016:779-788.
[5]	TAYEBI R M, MU Y, DEHKHARGHANIAN T, et al. Automated bone marrow cytology using deep learning to generate a histogram of cell types[J]. Commun Med, 2022,2:45.
[6]	MATEK C, KRAPPE S, MÜNZENMAYER C, et al. Highly accurate differentiation of bone marrow cell morphologies using deep neural networks on a large image data set[J]. Blood, 2021, 138(20):1917-1927. doi: 10.1182/blood.2020010568 pmid: 34792573
[7]	XIE S, GIRSHICK R, DOLLAR P, et al. Aggregated residual transformations for deep neural networks[C]. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE,2017: 5987-5995.
[8]	WANG Y, LIU H, WANG H, et al. Enhancing morphological analysis of peripheral blood cells in chronic lymphocytic leukemia with an artificial intelligence-based tool[J]. Leuk Res, 2023,130:107310.
[9]	OSMAN M, AKKUS Z, JEVREMOVIC D, et al. Classification of monocytes, promonocytes and monoblasts using deep neural network models: An area of unmet need in diagnostic hematopathology[J]. J Clin Med, 2021, 10(11):2264. doi: 10.3390/jcm10112264 URL
[10]	CAICEDO J C, MCQUIN C, GOODMAN A, et al. Weakly supervised learning of single-cell feature embeddings[C]. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. IEEE,2018: 9309-9318.
[11]	MANESCU P, NARAYANAN P, BENDKOWSKI C, et al. Detection of acute promyelocytic leukemia in peripheral blood and bone marrow with annotation-free deep learning[J]. Sci Rep, 2023,13:2562.
[12]	MA J, GUO Z, ZHOU F, et al. A generalizable pathology foundation model using a unified knowledge distillation pretraining framework[J]. Nat Biomed Eng, 2025.
[13]	WANG R, SHI Z, ZHANG Y, et al. Development and validation of deep learning model for diagnosis and subtypes differentiation of myeloproliferative neoplasms using clinical data and digital pathology[J]. Blood, 2023, 142(Supplement 1):123. doi: 10.1182/blood-2023-184459 URL
[14]	BRÜCK O E, LALLUKKA-BRÜCK S E, HOHTARI H R, et al. Machine learning of bone marrow histopathology identifies genetic and clinical determinants in patients with MDS[J]. Blood Cancer Discov, 2021, 2(3):238-249. doi: 10.1158/2643-3230.BCD-20-0162 URL
[15]	VAN EEKELEN L, PINCKAERS H, VAN DEN BRAND M, et al. Using deep learning for quantification of cellula-rity and cell lineages in bone marrow biopsies and comparison to normal age-related variation[J]. Pathology, 2022, 54(3):318-327. doi: 10.1016/j.pathol.2021.07.011 URL
[16]	LEE L H, MANSOOR A, WOOD B, et al. Performance of CellaVision DM96 in leukocyte classification[J]. J Pathol Inform, 2013, 4(1):14.
[17]	REZAEIKHONAKDAR D. AI chatbots and challenges of HIPAA compliance for AI developers and vendors[J]. J Law Med Ethics, 2023, 51(4):988-995. doi: 10.1017/jme.2024.15 URL
[18]	MINSSEN T, SOLAIMAN B, KÖTTERING L, et al. Gover-ning AI in the European Union: emerging infrastructures and regulatory ecosystems in health[M]//SOLAIMAN B, COHEN I G. Research handbook on health, AI and the Law Cheltenham. UK: Edward Elgar Publishing Ltd, 2024,311-331.
[19]	WANG X, ZHAO J, MAROSTICA E, et al. A pathology foundation model for cancer diagnosis and prognosis prediction[J]. Nature, 2024, 634(8035):970-978. doi: 10.1038/s41586-024-07894-z

人工智能在血液疾病形态学诊断中的应用进展

Application progress of artificial intelligence in morphological diagnosis of blood diseases

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献 19

相关文章 15

编辑推荐

Metrics

本文评价

[1]	李妙慧, 顾峻豪, 黄婷, 乔宇, 陈冰. 医学知识嵌入的因果框架在急性白血病骨髓形态学AI辅助诊断中发挥关键作用[J]. 诊断学理论与实践, 2026, 25(01): 44-52.
[2]	杨蕊馨, 于颖彦. 人工智能在消化道肿瘤医学图像数据处理的应用[J]. 诊断学理论与实践, 2025, 24(06): 605-612.
[3]	戴敏, 蒋裕雄, 史玉玲. 2023年至2024年银屑病诊治进展[J]. 诊断学理论与实践, 2025, 24(06): 593-604.
[4]	刘淦, 戴媛媛, 常文娇, 马筱玲. 脓毒症筛查技术进展[J]. 诊断学理论与实践, 2025, 24(06): 567-575.
[5]	余梦思, 宁聪华, 孟存仁, 帕丽扎提.阿不都热依木, 钟笛, 时瑛, 薛黎. 人工智能阅片平台在医学检验专业实习生血细胞形态教学中的应用研究[J]. 诊断学理论与实践, 2025, 24(06): 668-672.
[6]	洪烨娜, 张宇, SHI Kuangyu, 李彪, 郭睿. 放射性核素诊疗一体化的若干问题及对策[J]. 诊断学理论与实践, 2025, 24(03): 263-267.
[7]	徐梦迪, 高峰, 朱剑, 陈蕾, 秦雨萌, 黄越, 唐银萍, 沙杰. 新型海绵胶囊联合人工智能细胞DNA检测在早期食管癌筛查中的价值[J]. 诊断学理论与实践, 2024, 23(06): 580-586.
[8]	李卓含, 黄新韵, 郭睿, 李彪. ¹⁸F-FDG PET/CT在滤泡性淋巴瘤诊断和预后评估中的研究进展[J]. 诊断学理论与实践, 2024, 23(04): 439-444.
[9]	吴娜明, 李军, 陶娟. 恶性黑色素瘤的诊断热点[J]. 诊断学理论与实践, 2023, 22(03): 215-220.
[10]	唐静仪, 余群, 刘军. 结合人工智能的结构影像分析对阿尔茨海默病的早期预测及精准诊断研究进展[J]. 诊断学理论与实践, 2022, 21(01): 12-17.
[11]	徐浩, 张治, 解学乾, 杨文艺, 刘少稳. 冠脉生理功能评估软件（DEEPVESSEL FFR）与有创FFR在评估冠脉缺血中的对比研究[J]. 诊断学理论与实践, 2021, 20(04): 384-390.
[12]	柏媚, 顾怡, 周景艺, 张灏旻, 刘强, 沈薇, 李敏. 播散型马尔尼菲篮状菌感染1例[J]. 诊断学理论与实践, 2018, 17(06): 715-717.
[13]	许晶晶, 张敏鸣. 人工智能机器学习方法在阿尔茨海默病中的应用现状[J]. 诊断学理论与实践, 2018, 17(04): 466-470.
[14]	李菁原, 卢兴国. 成熟B细胞淋巴瘤的复杂性与细胞形态学[J]. 诊断学理论与实践, 2017, 16(05): 557-560.
[15]	黄文强, 范宇平, 胡烨, 杨阳, 江宁东, 黄媚媛, 潘家坪, 滕晓明. 精子形态、顶体区大小与IVF受精率的关系[J]. 诊断学理论与实践, 2017, 16(03): 270-272.