rxhands-unam-colab 0.24

Labeling of finger landmarks in hand xrays.

Finger landmark labeling for left handed x-ray images

This project uses two algorithms (symbolic and neural) to extract finger landmarks from left hand x-rays. It's intended to be used as an auxiliary tool for forensics research at UNAM.

Getting Started

The latest stable version can be downloaded from the PyPI.

Prerequisites

The code is written in Python 3, and it relies on OpenCV, SciPy and scikit-image:

• OpenCV
• SciPy
• scikit-image
• TensorFlow

Dependencies are automatically managed by pip.

Installing

To download, you can simply create a virtualenv and install the project with pip:

pip install rxhands-unam-colab

Command-line execution

Extraction can be performed with either the symbolic algorithm or the neural one.

rxhands [-h] -al {symbolic,neural} [-ch] [-pre] [-of OUTPUT_FOLDER] INPUT_FOLDER

where:

• -h help
• INPUT_FOLDER the path to the folder where the input images are stored.
• OUTPUT_FOLDER the path to the folder where the results will be stored.
• -al can receive either symbolic or neural.
• -pre means that gray level normalization will be applied to the input images.
• -ch if present, the script will try to crop the image to contain only the hand.

Examples

rxhands -al neural -of results/ data/

This command will label landmarks through the neural algorithm.

rxhands -al neural -ch -of results/ data/

This command will label landmarks through the neural algorithm. The images in folder data/ will be cropped on input.

rxhands -al neural -ch -pre -of results/ data/

This command will label landmarks through the neural algorithm. The images in folder data/ will be cropped and preprocessed on input.

rxhands -al symbolic -of results/ data/

This command will label landmarks through the symbolic algorithm.

Note that the symbolic algorithm only approximates landmarks in four fingers (not including the metacarpophalangeal joints).

Authors

• Arturo Curiel - Initial work - website

See also the list of contributors who participated in this project.

License

This project is licensed under the GNU/GPL3 License - see the LICENSE.md file for details

Acknowledgments

The neural model was implemented based on the architecture proposed by:

@inproceedings{Payer2016,
  title     = {Regressing Heatmaps for Multiple Landmark Localization Using {CNNs}},
  author    = {Payer, Christian and {\v{S}}tern, Darko and Bischof, Horst and Urschler, Martin},
  booktitle = {Medical Image Computing and Computer-Assisted Intervention - {MICCAI} 2016},
  doi       = {10.1007/978-3-319-46723-8_27},
  pages     = {230--238},
  year      = {2016},
}

Using the Digital Hand Atlas as training data.