DIABETIC RETINOPATHY DETECTION USING TRANSFER LEARNING AND DEEP CONVOLUTIONAL NEURAL NETWORKS

  • Unique Paper ID: 167902
  • Volume: 11
  • Issue: 4
  • PageNo: 1583-1590
  • Abstract:
  • The early detection and diagnosis of Diabetic Retinopathy (DR) can help prevent blindness in diabetic patients. In recent years, deep learning methods, specifically Convolutional Neural Networks (CNNs), have shown remarkable performance in detecting DR. In this study, we proposed the use of Densenet-201, a popular CNN architecture having 201 layers for diabetic retinopathy detection. We fine-tuned the pre-trained Densenet model on the dataset collected by Basys.ai, a healthcare company working towards the diagnosis of Diabetic retinopathy. Data augmentation techniques were used to make the data balanced since the data provided has imbalanced. Our results demonstrate that the Densenet model achieved high accuracy as compared to other state-of-the-art transfer learning models. Our model performed better than other published results using Densenet. We achieved a test accuracy of 92.50% and recall of 91.07% with precision 87.93%. The proposed approach holds great potential in assisting ophthalmologists in diagnosing DR and providing early interventions to patients.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 11
  • Issue: 4
  • PageNo: 1583-1590

DIABETIC RETINOPATHY DETECTION USING TRANSFER LEARNING AND DEEP CONVOLUTIONAL NEURAL NETWORKS

Related Articles