Soft Exudate Detection In Retinal Fundus Images

This application is designed to detect soft exudates (SEx) in retinal fundus images using image processing algorithms combined with machine learning. All the code for this project is available on my GitHub, and a detailed report can be found in the projects section of this website.

How to Use the Application

• Upload your fundus image in the designated area.
• Click Send to run the program.
• Wait for the result image (the analysis may take a few minutes).

You can also use the provided test image to see how the application works by selecting Use Test Image.

Image Requirements

• The image must be a fundus image, centered on the eye with a black background.
• It should not contain any prior segmentations or annotations.
• The image must be in color.
• Minimum resolution: 1500 x 1500 pixels.

Output

• Initial candidates identified by a basic algorithm, contoured in white.
• Suspicious SEx candidates contoured in blue.
• The optical disc contoured in black.
• Number of candidates identified by a basic algorithm.
• Number of Suspicious SEx candidates.

Example

Important Notice

This application is intended for academic purposes only and is not validated for use by medical professionals. The results may not be fully accurate, as the algorithm is still under development.