Glaucoma Detection Using U-Net and Multiple Localization

Intoduction:

Glaucoma is the second cause of blindness. Unfortunately, Indonesian people still unaware for the importance of early glaucoma detection. Moreover, Glaucoma Detection is the laborous and subjective job for ophtalmologists. Thus, we need the automation for glaucoma detection. We use glaucoma detection based on Optic Cup and Disc segmentation. This automation allow us to detect glaucoma more faster and less subjective. Furthermore, this system could be implemented in smartphone, hence more patient could monitor their glaucoma stage in much more affordable way. For further algorithm explanation, you can read our paper here.

The main algorithm for our Glaucoma Detection is follow:

1. Preprocessing
2. Optic Disc Localization
3. Optic Disc dan Cup Segmentation
4. Glaucoma Feature Extraction
5. Glaucoma Prediction

Usage

Below are the procedure for using this script:

1. Open your command line interface (command prompt, powershell, etc.) and go to project directory.
2. Install all libraries in requirements.txt (pip install -r requirements.txt).
3. Run this: python main.py /path/to/retinal_image.jpg in command prompt. Make sure you are in 'Code' directory.
4. After a few seconds, the detection result should appear in your command line interface and segmentation result would appear from matplotlib window.

Dataset link:

1. Drishti-GS1
2. REFUGE (probably you have to ask miccai 2020 permission to download the dataset)

Notebooks:

Main Notebook:

Localization Notebook:

Segmentation Notebook:

Glaucoma prediction Notebook:

Reference

A. N. Almustofa, A. Handayani, and T. L. R. Mengko, "Optic Disc and Optic Cup Segmentation on Retinal Image Based on Multimap Localization and U-Net Convolutional Neural Network," Journal of Image and Graphics, Vol. 10, No. 3, pp. 109-115, September 2022. link