深度学习在抗核抗体检测应用的进展及挑战

Application of deep learning in antinuclear antibodies classification: progress and challenges

抗核抗体(ANA)的实验室检测对系统性自身免疫病的诊断、分型、病情监测等具有重要的临床意义。近年来,随着计算能力的增强和算法的推陈出新,以深度学习(DL)为代表的人工智能技术取得不断突破,在医学图像识别领域逐渐展现出独特的优势。ANA检测的参考方法是以人喉癌上皮细胞为基质的间接免疫荧光法,检测结果依赖肉眼对荧光模式的判读,本质原理还是图像识别,而这正巧具备与DL结合实现自动化判读系统的广阔前景。本文就目前DL运用在ANA检测的领域相关的研究及面临挑战进行概述,以期为今后ANA结果判读的标准化之路提供参考。

Antinuclear antibodies(ANA)testing is essential for the diagnosis,classification,and disease activity monitoring of systemic autoimmune rheumatic diseases.In recent years,with the enhancement of computing power and the innovation of algorithms,the newly hip branch,deep learning(DL),practically delivered all of the most stunning achievements and breakthroughs in artificial intelligence(AI)so far.The application of DL to visual tasks,known as computer vision,has revealed significant power within the medical image recognition.Indirect immunofluorescence on HEp-2 cells is the reference method for ANA testing,the results is interpreted manually by specialized physicians.ANA fluorescent pattern classification is based on image recognition,which has a broad prospect of combining with DL to realize automatic interpretation system.This paper reviews the recent research progress and challenges of DL in the field of ANA detection in order to provide references for the standardization of ANA testing in the future.