Related Experiment Video
Updated: Oct 11, 2025
07:41
Behavioral Assessment of Visual Function via Optomotor Response and Cognitive Function via Y-Maze in Diabetic Rats
Published on: October 23, 2020
6.1K
Deep long and short term memory based Red Fox optimization algorithm for diabetic retinopathy detection and
Raju Pugal Priya1, Thankamony Saradadevi Sivarani2, Athimoolam Gnana Saravanan3
1Department of Electronics and Communication Engineering, Arunachala College of Engineering for Women, Kanyakumari, India.
Summary
A new deep learning algorithm, deep long- and short-term memory with Red Fox optimization (deep LSTM-RFO), accurately classifies diabetic retinopathy (DR). This method addresses limitations in existing DR detection, offering improved performance for early vision loss prevention.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Diabetic retinopathy (DR) is a leading cause of vision loss in diabetic patients.
- Current DR diagnosis faces challenges including high computational costs and difficulty tuning hyperparameters.
- Existing methods often fail to adequately reduce feature dimensions and increase execution times.
Purpose of the Study:
- To propose an efficient deep learning algorithm for accurate diabetic retinopathy classification.
- To overcome the limitations of existing DR detection methods.
- To improve the accuracy and efficiency of DR diagnosis.
Main Methods:
- Developed a deep long- and short-term memory (LSTM) neural network integrated with Red Fox optimization (deep LSTM-RFO).
- Employed a four-stage process: fundus image preprocessing (adaptive histogram equalization), lesion segmentation (adaptive watershed), feature extraction (statistical, intensity, color, shape), and classification.
- Utilized MESSIDOR, STARE, and DRIVE datasets for training and validation in MATLAB.
Main Results:
- The deep LSTM-RFO algorithm achieved high classification performance.
- Achieved 98.45% specificity, 96.78% sensitivity, 97.92% precision, 96.89% recall, and 97.93% F-score.
- Demonstrated superior results compared to previous methods in DR classification.
Conclusions:
- The proposed deep LSTM-RFO algorithm offers a robust and accurate solution for diabetic retinopathy classification.
- This method effectively addresses computational complexities and hyper-parameter tuning issues.
- The findings suggest a promising approach for early detection and prevention of vision loss due to DR.
