基于深度学习的糖尿病视网膜病变诊断方法研究

Research on the Detection Method of Diabetic Retinopathy Based on Deep Learning

摘要: 由于计算机技术以及人工智能的崛起和高速发展完善，图像处理和计算机视觉一类的高新技术开始被应用到医疗诊断的领域。传统对糖尿病视网膜病变(糖网病)的诊断受到医疗资源分布不均、医生的个人主观因素等影响较大，并且诊断过程耗时较长，本项目是以研究基于深度学习神经网络的方法来预测视网膜是否有病变，或者有轻度、高度、重度、增值型病。预期通过研究经典的神经网络模型，寻找一种基于深度学习智能算法的糖网病病变程度诊断方法，并搭建本地PC端可视化界面，使其具有易操作、普及度高的优势，实现对糖尿病视网膜病变的初步预测，辅助医生进行诊断治疗。

Abstract: Due to the rise and rapid development of computer technology and artificial intelligence, new technologies such as image processing and computer vision are beginning to be applied to the field of medical diagnosis. Traditional diagnostic methods of diabetic retinopathy (DR) are greatly affected by the uneven distribution of medical resources and doctors’ personal subjective factors. In addition, the diagnostic process takes a long time. Based on the deep learning and neural network, this paper searches for the best way to predict the presence of lesions in the retina or mild, high-grade, severe, or value-added disease. By studying the classic neural network model, we look for a diagnostic method for the degree of diabetic retinopathy based on deep learning intelligent algorithm, and build a local PC-side visualization interface and physical display module, so that it has the advantages of easy operation and high popularity, realizes the preliminary prediction of diabetic retinopathy, and assists doctors in diagnosis and treatment.