This application leverages AI to analyze medical images, specifically brain images and chest X-rays. Users can upload medical images, get predictions about potential issues, and verify or correct these predictions. The application ensures a smooth and secure user experience through a robust authentication system.
- User Authentication: Secure login system to protect user data.
- Medical Image Analysis: Upload and analyze brain images and chest X-rays.
- AI Predictions: Get AI-generated predictions for uploaded images.
- User Verification: Verify or correct AI predictions to improve accuracy.
- Responsive Design: User-friendly interface
- Secure login page powered by Asgardeo for user authentication.
- The home page allows users to select between chest X-ray analysis and brain image analysis.
- After login, users can choose between different types of medical scans to analyze.
- Upload brain images for AI analysis. The AI predicts the presence of tumors, and users can verify or correct these predictions.
- Upload chest X-rays for AI analysis. The AI detects issues such as pneumonia and provides a detailed analysis.
- Users can interact with the system to get more detailed explanations of the findings with chat.
- Users can correct AI predictions, enhancing the learning and accuracy of the model.
To run this application locally, follow these steps:
-
Clone the repository:
git clone https://github.com/ThemiraChathumina/medAI.git cd medAI npm install
-
Update the config.ts in the root directory and add your Asgardeo configuration details::
const baseURL = "http://localhost:5173"; export const config = { signInRedirectURL: baseURL, signOutRedirectURL: `${baseURL}/login`, clientID: "YOUR_CLIENT_ID", baseUrl: "https://api.asgardeo.io/t/YOUR_TENANT_NAME", scope: ["openid", "profile"], };
-
Setup and run the backend in medAI Backend.
-
Run the application:
npm run dev
This application uses models and techniques described in the following paper:
Interactive and Explainable Region-guided Radiology Report Generation
Tim Tanida, Philip Müller, Georgios Kaissis, Daniel Rueckert
In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023.
The code for the models can be found in their GitHub repository.
The following datasets were used for model training: