基于CTCNet对球结膜进行糖尿病视网膜病变分类

糖尿病视网膜病变(diabetic retinopathy, DR)是一种糖尿病性微血管病变，会在球结膜微血管上有所体现，球结膜血管图像的获取比眼底图像更加便捷，但微血管的特征变化微小且难以量化。为了能够对患者进行早期辅助诊断，本文依据球结膜微血管形态与DR的关联，首先对球结膜图像进行预处理，使用限制对比度自适应直方图均衡(contrast limited adaptive histogram equalization, CLAHE)算法进行图像增强，随机处理使数据增强，然后结合卷积神经网络(convolutional neural network, CNN)和Transformer各自的网络优势构建CTCNet,对处理后的球结膜血管图像进行DR分类，分类准确率达到了97.44%,敏感度97.69%,特异性97.11%,精确度97.69%,通过实验对比CNN和Transformer, CTCNet网络性能优于其他模型，能够有效识别DR。

Classification of diabetic retinopathy in bulbar conjunctiva based on CTCNet

Diabetes retinopathy (DR) is a kind of diabetes microvascular disease, and it will be reflected in the bulbar conjunctival microvessels. Images of bulbar conjunctival vessels are easier to obtain than fundus images,but the characteristic changes of microvessels are subtle and difficult to quantify. In order to enable early auxiliary diagnosis of patients,according to the association between the morphology of bulbar conjunctiva microvessels and DR, this paper first preprocesses the bulbar conjunctiva images,enhances the image using the contrast limited adaptive histogram equalization (CLAHE) algorithm,and enhances the data with random processing.Then CTCNet is constructed by combining the advantages of convolutional neural network (CNN) and Transformer, and the processed images of bulbar conjunctival vessels are classified into DR.The classification accuracy reaches 97.44%,sensitivity 97.69%,specificity 97.11%,and accuracy 97.69%.Through experimental comparison between CNN and Transformer,CTCNet has better performance than other models and can effectively identify DR.