AILungMeasure 0.1.5

Automated lung size measurements using deep learning and computer vision on portable chest radiographs.

AILungMeasure

AILungMeasure is a Python package designed for automated lung size measurements using deep learning and computer vision on portable chest radiographs. This package aims to improve accuracy, reduce variability, and streamline the lung transplantation size matching process.

Features

• Automated lung mask extraction from chest radiographs
• Feature points detection to generate lung height and width measurements
• Validation against measurements reported by radiologists
• Robust performance even with technically challenging radiographs

Installation

You can install AILungMeasure via pip:

pip install AILungMeasure

Usage

Loading the Model

AILungMeasure includes a pre-trained model for lung size measurements. When importing the package, the model will be automalically downloaded and loaded.

import AILungMeasure

Segmenting Lung Images

To get the lung mask, use the get_mask function:

input_image = '/path/to/your/image' # can be dicom, jpg, png, etc.. 
mask = AILungMeasure.get_mask(input_image)

To visualize the mask, use the show_mask function:

mask = AILungMeasure.show_mask(input_image)
 
# or plot it 
matplotlib.pyplot.imshow(mask)

Get Measurements

To plot the lung measurements, use the show_measurments function:

AILungMeasure.show_measurments(input_image , dpi=200) # dpi controls the image resolution

To get the measurments, use the get_measurments function:

measurments = AILungMeasure.get_measurments(input_image)

Note that this will work if the input_image is dicom, which already includes the pixel size in its meta data. The reported measurments will be in mm.
If you are using jpg, png, etc.. then the returned measurments will be in pixels.

License

This project is licensed under the Creative Commons ATTRIBUTION-NONCOMMERCIAL-SHAREALIKE 4.0 INTERNATIONAL - see the LICENSE file for details.

Contact

For questions or issues, please contact Mostafa Ismail at [e-mail].