机器学习对乳腺癌诊断应用现状研究进展  

Research advances in the application of machine learning in breast cancer diagnosis

作　　者：卢峰 张开炯[3] 吴立春 蒋叙川[2] 冀承杰 刘靳波 LU Feng;ZHANG Kaijiong;WU Lichun;JIANG XuchuanJI Chengjie;LIU Jinbo(Department of Laboratory Medicine,the Af filiated Hospital of Southwest Medical University,Luzhou,Sichuan646000,China;Department of Experimental Medicine,the People's Hospital of Jianyang City,Jianyang,Sichuan 641400,China;Department of Medical Laboratory,Sichuan Cancer Hospital Institute,Sichuan Cancer Prevention Center,the Af filiated Tumor Hospital of University of Electronic Science and Technology,Chengdu,Sichuan 610oo0,China)

机构地区：[1]西南医科大学附属医院医学检验部,四川泸州646000 [2]简阳市人民医院实验医学科,四川简阳641400 [3]四川省肿瘤医院·研究所,四川省癌症防治中心,电子科技大学附属肿瘤医院检验科,四川成都610000

出　　处：《社区医学杂志》2023年第24期1315-1322,共8页Journal Of Community Medicine

基　　金：四川省科技厅重点研发项目第二版(2022YFS0335);成都市医学科研课题(2022417);简阳市人民医院科研课题(JY202234)。

摘　　要：乳腺癌已经超越肺癌成为世界第一大癌症,同时也是女性患者中致死率最高的恶性肿瘤。随着乳腺癌诊疗技术的不断进步,早期乳腺癌患者的5年生存率可达95%,因此,乳腺癌早期筛查、诊断和治疗是其良好预后的关键。机器学习是人工智能(AI)领域极其重要的分支,由于计算机硬件的不断提升、深度学习算法的持续进步及海量乳腺癌临床数据的累积,使得机器学习在乳腺癌早期筛查、诊断及预后判断等领域发挥重要作用。本综述对机器学习在乳腺癌诊断领域中的研究及应用现状进行总结,以期为乳腺癌AI诊断提供新的方向和思路。以“机器学习”“深度学习”“人工智能”“乳腺癌诊断”为关键词,检索2016-01-01-2022-12-31 PubMed及中国知网相关文献。纳入标准:机器学习在乳腺癌X射线片诊断中的研究;机器学习在乳腺癌MRI诊断中的研究;机器学习在乳腺癌超声诊断中的研究;机器学习在乳腺癌病理诊断中的研究。排除标准:涉及乳腺癌AI诊断的著作、汇编及二次文献。最终纳入分析文献66篇。结果表明,机器学习在乳腺X射线片和乳腺超声领域的研究可以做到自动分割病灶、病灶特征的提取和分析,最终判别病灶的良恶性;而机器学习在乳腺MRI和乳腺病理诊断中的应用可以减轻临床医师的工作负担,提高诊断效率,还能最大限度的弱化临床医师诊断主观性和不稳定性。随着医疗数据的不断积累和AI技术飞速发展,医师在诊疗活动中可能成为新角色“AI医师”。未来的研究将会推动多中心高质量的大型公共数据库建立,在大数据基础上进行小数据的精细化标注可以提升乳腺癌诊断模型的泛化能力。另外,运用乳腺X射线、MRI、超声及病理多模态联合诊断可以进一步提升乳腺癌的诊断性能。Breast cancer has surpassed lung cancer to become the largest cancer in the world,and it is also the malignant tumor with the highest mortality rate among female patients.With the continuous progress of breast cancer diagnosis and treatment technology,the five-year survival rate of patients with early breast cancer can reach 95%.Therefore,early screening,diagnosis and treatment of breast cancer are the key to their good prognosis.Machine learning is an extremely important branch of artificial intlligence(AI).Due to the continuous improvement of computer hardware,the continuous progress of deep learning algorithms and the accumulation of massive clinical data of breast cancer,machine learning could play an important role in the fields of early screening,diagnosis and prognosis judgment of breast cancer.This study summarizes the research and application status of machine learning in the field of breast cancer diagnosis,in order to provide new directions and ideas for AI diagnosis of breast cancer.With machine learning,deep learning,artificial intelligence,and January 1,2016 to December 31,2022 were searched.Inclusion criteria:Study of machine learning in X-ray diagnosis of breast cancer;Study on machine learning in MRI diagnosis of breast cancer:Study on machine learning in ultrasonic diagnosis of breast cancer;Study on machine learning in pathological diagnosis of breast cancer.Exclusion criteria:works,compilations and secondary documents related to AI diagnosis of breast cancer.Finally,66 articles were included in the analysis.The results show that the research of machine learning in the field of mammography and breast ultrasound can automatically segment the lesions,extract and analyze the characteristics of the lesions,and finally distinguish the benign and malignant lesions.The application of machine learning in breast MRI and breast pathological diagnosis can reduce the workload of clinicians,improve the diagnostic efficiency,and weaken the subjectivity and instability of clinicians diagnosis to the maximum e

关 键 词：乳腺癌 人工智能 机器学习 深度学习 综述文献 

分 类 号：R737.9[医药卫生—肿瘤]