Scholarlink

Diabetic Retinopathy is a key factor in blindness worldwide, underlining the importance of its prior along with accurate discovery for the prevention of serious consequences. This work presents the suitability of ResNet-50 model for the categorization of diabetic retinopathy, attaining the high value accuracy of 97.56%. This model is observed for proving its efficiency for the retina scan analysis, correctly labeling the different stages of the disorder, and being very promising for the facilitation of the process of clinical decision-making. This achievement can improve patient treatment by making the intervention earlier possible and lessening the diagnostic load on medical professionals. The analysis also showcases capacity for transformation by using automated processing platform for medical images, leading path for increased efficiency for ophthalmic diagnostics. Architecture optimization for real-time operation, the use of multimodal clinical data, and greater explainability by using increased visualizations will also make its utility even greater clinically. This study is indicative of the high potential for AI methodology like ResNet-50 for the transformation of diabetic retinopathy screening progress of the medical domain overall.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Convolutional Neural Network with Attention for Multi-stage Diabetic Retinopathy Detection

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jigyasha Bhushan,
  • Gaurav Payal,
  • Geetika Verma,
  • Kanishk Rawat,
  • Bharti Chugh

摘要

Diabetic Retinopathy is a key factor in blindness worldwide, underlining the importance of its prior along with accurate discovery for the prevention of serious consequences. This work presents the suitability of ResNet-50 model for the categorization of diabetic retinopathy, attaining the high value accuracy of 97.56%. This model is observed for proving its efficiency for the retina scan analysis, correctly labeling the different stages of the disorder, and being very promising for the facilitation of the process of clinical decision-making. This achievement can improve patient treatment by making the intervention earlier possible and lessening the diagnostic load on medical professionals. The analysis also showcases capacity for transformation by using automated processing platform for medical images, leading path for increased efficiency for ophthalmic diagnostics. Architecture optimization for real-time operation, the use of multimodal clinical data, and greater explainability by using increased visualizations will also make its utility even greater clinically. This study is indicative of the high potential for AI methodology like ResNet-50 for the transformation of diabetic retinopathy screening progress of the medical domain overall.