Eye Disease Prediction

Introduction

This project utilizes a dataset of 712 ocular staining images to classify and predict eye diseases based on three main attributes: Category, Type, and Grade. The dataset encompasses a variety of corneal ulcer presentations, aiding in accurate diagnosis and treatment decisions.

Dataset

The dataset contains images labeled with the following categories, types, and grades of eye diseases:

Categories:

• Category 0: Point-like corneal ulcers
• Category 1: Point-flaky mixed corneal ulcers
• Category 2: Flaky corneal ulcers

Types:

• Type 0: No ulcer of the corneal epithelium
• Type 1: Micro punctate
• Type 2: Macro punctate
• Type 3: Coalescent macro punctate
• Type 4: Patch (≥1 mm)

Grades:

• Grade 0: No ulcer of the corneal epithelium
• Grade 1: Corneal ulcers involve only one surrounding quadrant
• Grade 2: Corneal ulcers involve two surrounding quadrants
• Grade 3: Corneal ulcers involve three or four surrounding quadrants
• Grade 4: Corneal ulcers involve the central optical zone of the cornea

Link to Project

https://www.kaggle.com/code/praths71018/eye-disease-detection

Acknowledgements

We acknowledge the SUSTech-SYSU dataset, curated by Deng et al., which provides a valuable resource for automatically segmenting and classifying corneal ulcers. For more details, refer to the original publication: Sci Data 7, 23 (2020).