Related Experiment Video
Updated: Sep 6, 2025
07:44
In vivo Structural Assessments of Ocular Disease in Rodent Models using Optical Coherence Tomography
Published on: July 24, 2020
3.0K
Deep Residual Network for Diagnosis of Retinal Diseases Using Optical Coherence Tomography Images
Sohaib Asif1, Kamran Amjad2, Qurrat-Ul-Ain3
1School of Computer Science and Engineering, Central South University, Changsha, Hunan, China. sohaibasif@csu.edu.cn.
Interdisciplinary Sciences, Computational Life Sciences
|June 29, 2022
Summary
A novel deep residual network accurately classifies four retinal diseases from Optical Coherence Tomography (OCT) images. This AI model achieves 99.48% accuracy, aiding early diagnosis and preventing blindness in diabetic retinopathy patients.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Diabetic retinopathy is a leading cause of blindness, progressing asymptomatically.
- Optical Coherence Tomography (OCT) is vital for diagnosing retinal conditions.
- Existing deep learning methods for OCT image analysis face reliability and interpretability challenges.
Purpose of the Study:
- To develop a reliable and interpretable deep learning model for classifying four retinal diseases in OCT images.
- To improve the accuracy and efficiency of automated diagnosis of diabetic macular edema (DME), choroidal neovascularization (CNV), DRUSEN, and normal OCT scans.
Main Methods:
- A modified ResNet50 deep residual network architecture was proposed.
- The model was pre-trained on ImageNet and trained end-to-end on a large dataset of 84,452 OCT images.
- A new fully connected block was introduced to enhance classification accuracy and prevent overfitting.
Main Results:
- The proposed model achieved an overall classification accuracy of 99.48% on a test set of 968 OCT images.
- It demonstrated superior performance compared to existing methods on the same dataset.
- The model achieved only 5 misclassifications, indicating high diagnostic precision.
Conclusions:
- The developed deep residual network is highly effective for diagnosing multiple retinal diseases from OCT images.
- The model shows significant promise for integration into ophthalmology clinics for early disease detection.
- This AI-driven approach can aid in preventing vision loss associated with diabetic retinopathy and other retinal conditions.
