人工智能血细胞形态学检查的关键技术进展  

Progress in key technologies of artificial intelligence-assisted blood cell morphology examination

作　　者：杨军霞[1] 连荷清 庞博[1] Yang Junxia;Lian Heqing;Pang Bo(Clinical Laboratory,Guang′anmen Hospital,China academy of Chinese medical sciences,Beijing 100053,China;Beijing Xiaoying Technology Co.,Ltd.,Beijing 100084,China)

机构地区：[1]中国中医科学院广安门医院检验科,北京100053 [2]北京小蝇科技有限责任公司,北京100084

出　　处：《中华检验医学杂志》2023年第3期326-330,共5页Chinese Journal of Laboratory Medicine

摘　　要：利用人工智能辅助血细胞形态学检查具有十分广阔的临床应用前景,它可以大幅提高工作效率,减轻人力负担,避免主观化,有利于实现标准化。其主要的重点和难点在于图像获取、图像分割、细胞识别与分类等几个关键的技术环节。近年来,无论是硬件设备还是软件算法都进展迅速,这使得人工智能辅助系统从数字图像采集、白细胞分割到细胞特征提取、分类等方面都出现了重要的发展。其中,相对于传统的机器学习而言,深度学习技术在血细胞形态学识别方面的应用尤其值得关注。此外,显微血细胞图像数据库的不断涌现也为各种算法技术的进一步发展和完善提供了重要支撑。了解人工智能辅助血细胞形态学检查的关键技术进展将有利于推进其不断发展,更好地走向临床应用。近年来人工智能技术从“传统的机器学习”向“深度学习”转变,不再依赖于人工提取特征,而是依靠其自动抽取数据的能力来实现。与国外血细胞图像数据库相比,国内数据库还有较大差距,应加强此方面的建设。Artificial intelligence-assisted blood cell morphology examination of blood cells is very promising in clinical applications.Because it can significantly improve work efficiency,reduce the burden of manpower,avoid subjectivism,and facilitate standardization.The main difficulties lie in several key technical links,such as image acquisition,image segmentation,cell identification,and classification,etc.In recent years,both hardware devices and software algorithms have made rapid progress,which has led to the important development of artificial intelligence auxiliary systems from digital image acquisition,white blood cell segmentation,cell feature extraction,and classification.Compared with the traditional machine learning,the application of deep learning technology in the morphological identification of blood cells is particularly worthy of attention.In addition,the continuous emergence of microscopic blood cell image databases also provides important support for the further development and improvement of various algorithms.Understanding the key technical progress of artificial intelligence-assisted blood cell morphology examination will help to promote its continuous development and better clinical application.In recent years,artificial intelligence technology has changed from"traditional machine learning"to"deep learning",which no longer relies on manual extraction of features,but on its ability to automatically extract data to achieve.Compared with the blood cell image database from foreign countries,the construction of domestic databases should be strengthened to minimize the gap between foreign databases.

关 键 词：白血病 人工智能 血细胞 形态学 

分 类 号：R446.113[医药卫生—诊断学]