AnemoScan: AI-Based Early Detection of Anemia from Nail and Conjunctiva Images
AnemoScan is an Artificial Intelligence-powered healthcare application designed to support the early screening of anemia through non-invasive image analysis. The system utilizes deep learning and computer vision techniques to analyze images of the conjunctiva (inner eyelid) and fingernails, two clinically relevant regions commonly associated with anemia-related pallor.
The project employs transfer learning using MobileNetV2 to classify images as Anemic or Non-Anemic. To improve usability and automation, a dedicated image type classifier was developed to automatically identify whether an uploaded image belongs to the conjunctiva or nail category and route it to the appropriate diagnostic model.
To enhance transparency and trustworthiness, Explainable AI (XAI) techniques were incorporated using Grad-CAM, allowing users to visualize the image regions that influenced the model's predictions. The system was deployed through an interactive Streamlit web application that provides real-time predictions, confidence scores, visual explanations, downloadable PDF reports, and educational nutritional recommendations related to anemia management.
Key Features:
1. Automatic image type detection (Conjunctiva or Nail)
2. AI-based anemia classification
3. Confidence-based prediction system
4. Explainable AI visualization using Grad-CAM
5. Professional PDF report generation
6. Nutritional guidance module
7. User-friendly web interface for real-time screening
This project demonstrates the application of Artificial Intelligence and Computer Vision in healthcare by providing an accessible, low-cost, and non-invasive approach to early anemia screening.
