About the Project

The hardware system employs a meticulously designed image processing pipeline to ensure the highest quality and accuracy of diabetic retinopathy detection

Process Overview

The process commences with the acquisition of retinal images, also referred to as fundus images, utilizing a specialized retinal lens attached to a mobile device. The retinal lens is designed to capture high-resolution images of the patient’s retina, ensuring critical details are preserved. The mobile device serves as a user-friendly interface, enabling seamless image acquisition by patients or healthcare

Image Enhancement Techniques

• Contrast Adjustments: Improving the overall contrast of the image.
• Noise Reduction: Reducing unwanted noise to enhance clarity.
• Image Sharpening: Enhancing the sharpness of the image to reveal critical details.
• Normalization: Ensuring consistent brightness and contrast across different images.

Data Transfer and Processing

The preprocessed retinal image is seamlessly transferred to the core computational unit of the system, the Raspberry Pi, via a secure and efficient data transfer

Algorithmic Analysis

• Convolutional Neural Network (CNN): Extracts complex features from the image, detecting intricate patterns such as microaneurysms, hemorrhages, and exudates.
• Long Short-Term Memory (LSTM): Captures sequential patterns and temporal content, providing a dynamic understanding of disease progress over time.
• Integration: Outputs from both CNN and LSTM networks are integrated to generate a comprehensive understanding of the retinal image’s characteristics.

Stage Classification

The system classifies diabetic retinopathy into specific stages based on the integrated analysis results:

• Mild
• Moderate
• Severe
• Proliferative Diabetic Retinopathy

Data Storage and Output

• Data Storage: Diagnostic results and relevant data are stored in a secure and scalable system, which can include a database or cloud storage.
• Output Screen: Displays diagnostic results to healthcare professionals or patients, including the classification and stages of diabetic retinopathy, providing valuable information for treatment and management.