Apppublic

steeltroops-ai/medilens

sourceHugging Facemitupdated 8mo agoView on Hugging Face

0likes

App README

MediLens AI - Medical Diagnostics Backend

Clinical-grade AI backend for medical image analysis and biosignal processing.

Features

—RetinaScan AI: Diabetic retinopathy detection from fundus images
—SpeechMD AI: Voice biomarker analysis for neurological assessment
—CardioPredict AI: ECG signal analysis for cardiac conditions
—ChestXplorer AI: Chest X-ray analysis for pulmonary conditions
—AI Explanations: Powered by Cerebras Llama 3.3 70B

Tech Stack

—Framework: FastAPI with async support
—ML: PyTorch, TensorFlow, OpenCV
—Audio: Parselmouth (Praat), librosa
—AI: Cerebras Cloud, AWS Polly
—Database: SQLite with SQLAlchemy

API Endpoints

Endpoint	Description
`/api/health`	Health check
`/api/retinal/analyze`	Retinal fundus analysis
`/api/speech/analyze`	Voice biomarker extraction
`/api/cardiology/analyze`	ECG signal analysis
`/api/radiology/analyze`	Chest X-ray analysis
`/api/explain`	AI explanation generation
`/api/voice`	Text-to-speech synthesis

Environment Variables

Variable	Description	Required
`CEREBRAS_API_KEY`	Cerebras Cloud API key	Yes
`AWS_ACCESS_KEY_ID`	AWS access key for Polly	Optional
`AWS_SECRET_ACCESS_KEY`	AWS secret key for Polly	Optional
`AWS_REGION`	AWS region (default: us-east-1)	Optional
`ELEVENLABS_API_KEY`	ElevenLabs API key	Optional

Running Locally

bash

cd backend
python -m venv .venv
source .venv/bin/activate  # or .venv\Scripts\activate on Windows
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

Docker

bash

docker build -t medilens-backend .
docker run -p 7860:7860 --env-file .env medilens-backend

Disclaimer

This is a research and educational tool. Always consult with qualified healthcare providers for medical decisions.

License

MIT License