aliiakbarkhan/eye-disease-detection-model

Eye Disease Detection and Classification using Deep Learning

This dataset contains a trained deep learning model designed to classify and detect common eye diseases from medical images. Using a convolutional neural network (CNN) built with Fastai and PyTorch, the model provides automated, high-accuracy predictions based on retinal imagery.

Overview

This project applies deep learning to the field of ophthalmology by leveraging a pre-trained ResNet18 architecture fine-tuned on a labeled dataset of eye images. The model is capable of distinguishing between healthy eyes and various pathological conditions, aiding in early detection and potential diagnosis.

Problem Statement

Manual diagnosis of eye diseases can be time-consuming and resource-intensive. This model provides a step toward automation by identifying eye diseases from image data with high accuracy, potentially assisting doctors or being integrated into pre-screening tools.

Model Details

• —Architecture: ResNet-18 (pretrained on ImageNet)
• —Framework: Fastai + PyTorch
• —Training Method: Fine-tuned for 10 epochs using vision_learner
• —Evaluation Metric: ~92% validation accuracy
• —Model Output: eye_disease_model.pkl

Eye Disease Categories

The model can classify eye images into the following categories:

• —Cataract
• —Glaucoma
• —Retina Disease
• —Normal

Author

Ali Akbar Khan GitHub: @aliiakbarkhan

References

• —Fastai Documentation
• —PyTorch
• —Eye Disease Datasets on Kaggle