Related Experiment Video
Updated: May 10, 2025
09:17
Using Retinal Imaging to Study Dementia
Published on: November 6, 2017
21.4K
Diabetic retinopathy detection based on mobile maxout network and weber local descriptor feature selection using
V Sheejakumari1, K Sundravadivelu2, S Pushparani3
1Department of Computer Science and Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences (SIMATS), Saveetha University, Chennai, Tamilnadu, 602 105, India.
Scientific Reports
|April 22, 2025
Summary
This study introduces a new hybrid network, the Mobile Maxout network (MM-Net), for faster and more accurate diabetic retinopathy (DR) detection from retinal images. MM-Net improves upon existing methods, offering a promising tool for early DR diagnosis.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Diabetic retinopathy (DR) detection from color fundus images requires expert analysis, which is time-consuming and complex.
- Current automated DR detection methods can be slow and costly, highlighting a need for improved efficiency.
Purpose of the Study:
- To develop an efficient and accurate hybrid network, Mobile Maxout network (MM-Net), for automated diabetic retinopathy detection.
- To overcome the limitations of manual DR grading and existing automated systems.
Main Methods:
- A hybrid MM-Net, combining MobileNet and Deep Maxout Network (DMN), was proposed for DR detection.
- Image preprocessing involved a median filter, followed by optic disk segmentation using an active contour model and blood vessel segmentation via O-SegNet.
- Feature extraction and final DR detection were performed using the MM-Net.
Main Results:
- The MM-Net achieved high analytical metrics for DR detection.
- Accuracy reached 89.2%, sensitivity was 90.5%, and specificity was 92.0%.
Conclusions:
- The proposed MM-Net demonstrates significant potential for accurate and efficient automated diabetic retinopathy detection.
- This hybrid network offers a viable solution to challenges in DR screening and diagnosis using retinal fundus images.
