Retinal Disease Detection for the Second largest Ophthalmology Clinic in Asia, backed by Crys Capital

Business Problem

 

  • Delayed Diagnosis: Manual retinal disease diagnosis is time-consuming and requires expert ophthalmologists. 
  • High Cost of Screening: Traditional diagnostic methods require specialized equipment and trained professionals, increasing healthcare costs. 
  • Lack of Accessibility: Many regions, especially rural areas, lack access to expert diagnosis and screening facilities. 
  • Data Imbalance and Overfitting: Many existing AI models suffer from biased datasets, leading to inaccurate predictions. 
  • Scalability Issues: Deploying a high-performance model for large-scale disease detection requires efficient computational management.  

About Retinal Image Extraction

Retinal diseases such as Cataract, Diabetic Retinopathy, and Glaucoma are leading causes of vision impairment worldwide. Early detection and classification of these diseases play a crucial role in timely medical intervention. This project leverages Convolutional Neural Networks (CNNs) to build an automated image classification system capable of identifying and categorizing retinal diseases into four classes: Cataract, Diabetic Retinopathy, Glaucoma, and Normal Retina. The solution involves training and deploying two CNN architectures, SmallRetinaNet and RetinaNet, to achieve efficient and accurate disease detection.  

Explore Now

Solution

To address these challenges, an advanced solution leveraging cutting-edge technologies was developed. The solution focuses on precision, scalability, and usability to empower healthcare providers and improve clinical outcomes. 

Architecture

  • User Interaction 
    The user interacts with the Application (Web/Mobile) to upload retinal images. 
  • Authentication & Access Control 
    AWS Cognito handles user authentication and access management. 
  • Content Delivery & API Management 
    The Application communicates via AWS CloudFront for optimized content delivery. 
    AWS API Gateway manages incoming API requests securely. 
  • Image Processing & Storage 
    AWS Lambda processes uploaded images. 
    Processed images and metadata are stored in AWS S3
  • Model Inference 
    The stored images are sent to AWS SageMaker, where the trained CNN model performs disease classification. 
  • Data Storage & Retrieval 
    Predictions and metadata are stored in AWS DynamoDB for further reference and analysis. 
    AWS Lambda retrieves stored predictions and returns results to the user. 
Retinal Disease Detection for the Second largest Ophthalmology Clinic in Asia, backed by Crys Capital
Outcomes

0%

Enhanced Diagnostic Accuracy: RetinaNet achieved high accuracy, precision, recall, and F1-score, ensuring reliable disease classification. 

0%

Improved Accessibility: SmallRetinaNet allowed deployment in low-resource environments, making AI-driven screening accessible in rural areas. 

0%

Scalability & Efficiency: The dual-model approach enabled both high-performance analysis in hospitals and quick screening in primary care centers.