Apppublic

Amar-nadh/medsage-tensorflow

sourceHugging Facemitupdated 4mo agoView on Hugging Face

0likes

App README

Chest X-ray Pneumonia Detection API

A FastAPI inference server that uses a MobileNetV2 (Keras) model to detect pneumonia from chest X-ray images, with Grad-CAM heatmap visualisation.

Endpoints

Method	Path	Description
`GET`	`/`	Health check
`GET`	`/health`	Detailed health status
`GET`	`/stats`	Model performance statistics
`POST`	`/predict/xray-pneumonia`	Predict pneumonia from an uploaded X-ray image

Usage

Predict

bash

curl -X POST \
  https://Amar-nadh-chest-xray-pneumonia-detection.hf.space/predict/xray-pneumonia \
  -F "file=@chest_xray.jpg"

Response

json

{
  "diagnosis": "PNEUMONIA",
  "confidence": 92.34,
  "confidence_level": "High",
  "recommendation": "Findings suggest pneumonia. Clinical correlation and follow-up recommended.",
  "raw_score": 0.923401,
  "heatmap_base64": "<base64 PNG>",
  "validation_metrics": {
    "accuracy": 86.0,
    "sensitivity": 96.4,
    "specificity": 74.8,
    "precision": 80.4,
    "roc_auc": 0.964
  }
}

Supported Formats

—Images: JPEG, PNG, WebP, BMP, TIFF
—Medical: DICOM (.dcm)

Model

—Source: `ayushirathour/chest-xray-pneumonia-detection`
—Architecture: MobileNetV2
—Input: 224 × 224 RGB
—Validated on: 485 independent samples

⚠️ Disclaimer

This tool is for educational and research purposes only. It is not a substitute for professional medical diagnosis.