多类视网膜疾病OCT图像分类方法研究  

Research on OCT Image Classification Methods for Multiple Types of Retinal Diseases

作　　者：徐国荣 Guorong Xu(School of Health Science and Engineering,University of Shanghai for Science and Technology,Shanghai)

机构地区：[1]上海理工大学健康科学与工程学院,上海

出　　处：《建模与仿真》2025年第1期1136-1145,共10页Modeling and Simulation

摘　　要：视网膜疾病是当前威胁人类视觉健康的重要问题之一,其早期诊断和干预对预防视力损害具有重要意义。光学相干断层扫描作为一种无创成像技术,在视网膜疾病诊断中发挥着关键作用。本文提出了一种基于改进DenseNet的深度学习模型,用于多类视网膜OCT图像的自动分类。在OCT-C8数据集上进行实验,结果表明改进后的模型在八类视网膜疾病的分类任务中表现优异,平均准确率达到99.41%。与现有其他方法相比,本文提出的模型展现出更优的分类性能。Retinal diseases are one of the most important problems threatening human visual health,and their early diagnosis and intervention are of great significance in preventing visual impairment.Optical coherence tomography,as a non-invasive imaging technique,plays a key role in the diagnosis of retinal diseases.In this paper,a deep learning model based on improved DenseNet is proposed for the automatic classification of multi-class retinal OCT images.Experiments on the OCT-C8 dataset show that the improved model performs well in the classification task of eight retinal diseases with an average accuracy of 99.41%.Compared with other existing methods,the proposed model shows better classification performance.

关 键 词：视网膜疾病 光学相干断层扫描 深度学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]