Airway Segmentation

3D Slicer extension for airway segmentation in CT/CBCT images.

The extension contains two modules:

• Airway Segmentation: This is a simple module that segments the airways from a CT/CBCT image. The user needs to only specify the input volume and a markup point placed in the airway lumen. The module automatically retrieves the convolution kernel of the image if the image is loaded from DICOM (otherwise STANDARD kernel is used), detects chest vs head/neck anatomy from DICOM metadata, and configures the CLI accordingly. The result is saved into a segmentation node. The module uses Airway Segmentation CLI module internally.
• Airway Segmentation CLI: CLI module that implements the segmentation algorithm. It uses a modified version of ITK's itkConnectedThresholdImageFilter to segment all pixels with intensity below an adaptive threshold. For CHEST anatomy it runs the original trachea + left/right branch pipeline. For UPPER_AIRWAY (head/neck CBCT), it segments only the connected upper-airway region from the provided seed and skips chest-specific carina/bronchial processing.

The repository was forked from https://github.com/PietroNardelli/Slicer-AirwaySegmentation because the maintainer did not merge pull requests for several years.

Tutorial

• Go to Sample Data module
• Click on CTChest to load the CT chest sample data set into the scene
• Go to Markups module
• Click + Point list button to create a new point list
• Click in the trachea in any slice view
• Go to Airway Segmentation module
• Select CTChest as CT volume
• Select F as Seed point
• Click Apply

User interface

• Inputs
  • CT volume: Input CT/CBCT dataset to be segmented.
  • Seed point: Seed point for the algorithm. Only one seed point must be placed in the airway lumen (trachea for chest CT, upper airway for head/neck CBCT).
• Outputs
  • Segmentation: Output segmentation. If left at default then a new segmentation is created automatically.

Related extensions

• Lung CT analyzer provides grow-cut algorithm based airway segmentation. See more information here.
• SlicerCIP (Chest Imaging Platform) extension contains a slightly improved version of this module (main difference that label value and voxel type is consistent with other modules of CIP and that it can extract the whole airway or trachea/left/right sides separately).

References

Nardelli, P., Khan, K. A., Corvò, A., Moore, N., Murphy, M. J., Twomey, M., O'Connor, O. J., Kennedy, M. P., Estépar, R. S. J., Maher, M. M. & Cantillon-Murphy, P. (2015). Optimizing parameters of an open-source airway segmentation algorithm using different CT images. Biomedical engineering online, 14(1), 62.

Acknowledgments

This work is supported by NA-MIC, the Slicer Community and University College of Cork.

• Author: Pietro Nardelli (University College Cork)
• Contributor: Andras Lasso (PerkLab, Queen's University)