tatic3/Chiremba-AI-Backend-Image-Diagnosis

Chiremba AI is an innovative health diagnosis system leveraging NLP for text-based diagnosis, CNNs for image-based disease detection (e.g., skin diseases), and telemedicine for virtual consultations. Designed for underserved communities in Zimbabwe, it provides accessible, affordable, and accurate healthcare solutions. # Chiremba AI Backend Image Diagnosis Service A comprehensive medical image analysis service that provides AI-powered diagnosis for multiple medical conditions. ## Features - Multi-model medical image analysis system - Support for multiple conditions: - Brain tumor detection and classification - Lung cancer detection - Pneumonia analysis - Skin disease classification - Advanced image preprocessing pipeline - High-performance API endpoints ## Technology Stack - FastAPI for high-performance API endpoints - TensorFlow 2.17 for model inference - OpenCV for image processing - Uvicorn for ASGI server ## Setup 1. Install dependencies: ```bash pip install -r requirements.txt ``` 2. Ensure model files are present in the `models/` directory: - `brain_tumor_model.h5` - `lung_cancer_model.h5` - `pneumonia_model.h5` 3. Start the service: ```bash python main.py ``` ## API Endpoints - `GET /`: Service information and status - `GET /health`: Health check endpoint - `POST /analyze/brain`: Brain tumor analysis endpoint - `POST /analyze/lung`: Lung cancer detection endpoint - `POST /analyze/pneumonia`: Pneumonia analysis endpoint All analysis endpoints accept: - Image file upload (DICOM, PNG, JPEG formats) - Optional parameters for analysis configuration ## Environment Variables - `PORT`: Server port (default: 8000) - `MODEL_PATH`: Custom path to model directory - `LOG_LEVEL`: Logging level (default: INFO) ## Image Processing Pipeline 1. Image validation and format checking 2. Preprocessing and normalization 3. Model-specific transformations 4. Inference and result aggregation 5. Confidence score calculation ## Dependencies See `requirements.txt` for full list of dependencies.