模式识别在眼底图像精准诊断分析系统中的应用研究

Research on the application of pattern recognition in accurate and precise diagnosis and analysis system of fundus image

目的:构建基于模式识别的眼底图像精准诊断分析系统,为临床提供眼底图像快速辅助分析手段。方法:采用Oracle数据库、Hadoop大数据存储技术、Spring Data数据访问组件、OpenCV图像分析方法库,构建眼底图像精准诊断分析系统,分别实现患者眼底图像采集、预处理、特征提取、特征检索、特征区域分割、病历信息关联、眼底图像特征库构建及辅助诊断等功能。选取在医院就诊的100例患者按照标签等比例将其分为训练样本(70例)和测试样本(30例),以眼底数字图像中黄斑疾病的辅助诊断为例进行测试。结果:眼底图像精准诊断分析系统为临床提供操作简便的智能化辅助诊断工具。经测试,在9例病例中检测出疑似黄斑分割结果,其中2例为假阳性结果,无假阴性结果。结论:构建眼部疾病图像精准诊断分析系统,可为眼部疾病诊断提供简便易行的应用工具,该方法具有较高的敏感度,在一定程度上可作为大范围筛查的辅助手段,可降低数字图像处理技术在临床的应用难度,为建立规范化培训体系提供数据支持。

Objective:To construct an accurate and precise diagnosis and analysis system of fundus image based on pattern recognition,which could provide a rapid auxiliary analysis method of fundus image for clinical work.Methods:The Oracle database,Hadoop big data storage technique,Spring Data data access component and OpenCV image analysis method base were adopted to build the accurate and precise diagnosis and analysis system of fundus image so as to respectively realize image acquisition,preprocessing,feature extraction,feature retrieval,segmentation of feature region,association of medical record information,construction of feature library of fundus image and auxiliary diagnosis for the fundus of patients.100 patients who admitted to hospital were selected and were divided into training samples(70 cases)and test sample(30 cases)as equal proportion according to their tag,and the auxiliary diagnosis of macula disease in the digital image of fundus was used as case to implement test.Results:The accurate and precise diagnosis and analysis system of fundus image could provide intelligent auxiliary diagnosis tool with simple operation for clinical work.And the tested results indicated that the segmentation results of suspected macula were found in 9 cases,and 2 cases of them were false positive,and there was no false negative result.Conclusion:The construction of accurate and precise diagnosis and analysis system of the image of oculopathy disease can provide a simple and easy application tool for the diagnosis of oculopathy.And it has higher sensitivity,and can be used as an auxiliary means of large-scale screening to a certain extent,and it can reduce the applied difficulty of digital picture processing technique in clinical application,and can provide data support for establishing standardized training system.