medical-image-classifier

API Reference

Base URL (local): http://localhost:8000 Interactive docs: http://localhost:8000/docs (Swagger UI) and /redoc.

All prediction responses include a disclaimer field. This service is for educational and portfolio demonstration only and is not for clinical use.

`GET /health`

Liveness/readiness probe.

Response 200

{ "status": "ok", "model_loaded": true }

`GET /model-info`

Metadata about the currently loaded model.

Response 200

{
  "architecture": "small_cnn",
  "classes": ["Cognitively Normal", "Mild Cognitive Impairment", "Alzheimer's Disease"],
  "image_size": 128,
  "num_parameters": 97779,
  "trained_on": "synthetic",
  "metrics": { "accuracy": 0.941, "macro_f1": 0.940, "dataset": "synthetic" },
  "created_at": "2025-01-01T00:00:00+00:00",
  "disclaimer": "..."
}

`POST /predict`

Classify a single medical image.

Request: multipart/form-data with a single file field named file. Accepted content types: PNG, JPEG, BMP, TIFF. Max size: 10 MB (configurable).

Example

curl -X POST http://localhost:8000/predict \
  -F "file=@samples/sample_ad_1.png"

Response 200

{
  "predicted_class": "Alzheimer's Disease",
  "confidence": 0.87,
  "class_probabilities": {
    "Cognitively Normal": 0.05,
    "Mild Cognitive Impairment": 0.08,
    "Alzheimer's Disease": 0.87
  },
  "input_metadata": {
    "filename": "sample_ad_1.png",
    "content_type": "image/png",
    "size_bytes": 11391
  },
  "disclaimer": "This model is for educational and portfolio demonstration only and is not intended for clinical diagnosis. Predictions are not medical advice and must not be used for clinical decision-making."
}

Error responses

Status	Meaning
`400`	Empty file body.
`413`	File exceeds the maximum upload size.
`415`	Unsupported content type.
`422`	File could not be decoded as an image.
`503`	No model is loaded (train one first).

`POST /explain`

Return a Grad-CAM heatmap overlay (PNG) highlighting the regions that most influenced the predicted class.

Request: same as /predict.

Example

curl -X POST http://localhost:8000/explain \
  -F "file=@samples/sample_ad_1.png" --output gradcam.png

Response 200: image/png binary body.

Configuration via environment variables

Any config value can be overridden with MIC_<SECTION>_<FIELD>:

Variable	Effect
`MIC_API_MODEL_PATH`	Path to the checkpoint to serve.
`MIC_API_PORT`	Port the app binds to (when using the config default).
`MIC_API_MAX_UPLOAD_BYTES`	Maximum accepted upload size.
`MIC_API_LOG_LEVEL`	Log verbosity (`DEBUG`, `INFO`, …).

This site is open source. Improve this page.