PyMAD

Python Medical Anomaly Detection (PyMAD) using k-NNN and visibility detection

This is an implementation of the k-NNN algorithm Anomaly detection in Medical Imaging as introduced in https://arxiv.org/abs/2305.17695 by Tal and Nizan.
The visibility detection code can be found in https://github.com/CV-Reimplementation/Document-Enhancement-using-Visibility-Detection?tab=readme-ov-file

Visibility detection pipeline

k-NNN pipeline (source)

Getting started

Prepare your images

Store train and test images using the following structure:

dataset_mame/
├── train/
│   ├── img1
│   ├── img2
│   ├── ...
├── test/
│   ├── good/
│     ├── img1
│     ├── img2
│     ├── ...
│   ├── ungood/
│     ├── img1
│     ├── img2
│     ├── ...

Running the algorithm

Main.py contains example code showing how to configure, train, and test the K-NNN model.

Results

To be updated soon