/feature-visualization

A toolkit for visualizing feature (heat)map by Pyradiomics.

Primary LanguagePython

Feature Visualization

uv

A toolkit for visualizing feature (heat)map by Pyradiomics.

asciicast

Pyradiomics is an excellent tool for feature extraction from medical imaging. However, it hasn't been born with built-in visualization tool. In our manuscript, a segment-based extraction mothod was involved, thus, a certain amount of features were extracted by volume of interest, each of which holds a single value.

The hidden meaning of these features can be obscure somehow. To settle this matter, we've developed this toolkit to provide intuitive feature heatmaps as image-level explanation. Both metioned segment-based method and this voxel-based share same strategy de facto.

Usage

One should have uv installed. Then clone the code to your local machine and tell the program you need more help:

git clone https://github.com/bioinformatist/feature-visualization.git
cd feature-visualization
uv run plot.py --help

Three subcommands were provided. They are:

  • list: List all available feature names according to provided PARAMS file.
  • plot: Plot the the feature map of MASKed IMAGE.
  • view: Save the IMAGE and MASK as tiff format named with OUTPUT.

Tip

All images will be rendered in .tiff format with 300 ppi resolution and lzw compressed (suitable for publication use).

Feel easy to submit a new issue if you encounter any problems.

Credits

This implementation depends heavily on zhangjingcode/RadiomicsFeatureVisualization.