Ohsaka Akimichi
Department of Transfusion Medicine and Stem Cell Regulation, Department of Next Generation Hematology Laboratory Medicine, Juntendo University Graduate School of Medicine.
Rinsho Ketsueki. 2020;61(5):564-569. doi: 10.11406/rinketsu.61.564.
Morphological analysis of the blood smear is an essential element of diagnosing a disease hematologically and has been performed by conventional manual light microscopy for several decades. Although this method is the gold standard, it is labor-intensive, requires continuous training of the personnel, and is subject to relatively large interobserver variability. The artificial intelligence (AI)-based automated methods for the digital morphological analysis of blood smears have recently been developed. In this review, our recently developed convolutional neural network (CNN)-based digital morphology hematology analysis system is introduced. AI-based digital morphology hematology analysis system is firstly needed to incorporate digital imaging of blood cells into the analysis system. It is essential to establish a digital platform, which was already established in the radiological diagnosis, for the dissemination of CNN-based automated digital morphology hematology analyzer in the near future.
血液涂片的形态学分析是血液学疾病诊断的重要组成部分，几十年来一直通过传统的手动光学显微镜进行。尽管这种方法是金标准，但它劳动强度大，需要对人员进行持续培训，并且观察者间的变异性相对较大。最近已经开发出基于人工智能（AI）的血液涂片数字形态学分析自动化方法。在这篇综述中，将介绍我们最近开发的基于卷积神经网络（CNN）的数字形态学血液学分析系统。基于AI的数字形态学血液学分析系统首先需要将血细胞的数字成像纳入分析系统。建立一个数字平台至关重要，这在放射诊断中已经实现，以便在不久的将来推广基于CNN的自动化数字形态学血液分析仪。
