Three-dimensional segmentation and skeletonization to build an airway tree data structure for small animals.

Area of Science:

• Pulmonary medicine
• Medical imaging
• Computational biology

Background:

• Quantitative analysis of intrathoracic airway tree geometry is crucial for evaluating bronchial tree structure and function.
• Existing morphometric data predominantly focuses on humans, with limited data available for small animal models.
• Accurate airway morphometry in small animals is essential for preclinical research and drug development.

Purpose of the Study:

• To develop and implement a semi-automatic computational approach for quantitative description of airway tree geometry in small animals.
• To build a tree data structure from high-resolution computed tomography (CT) images for small animal airway analysis.
• To address the gap in small animal airway morphometric data.

Main Methods:

• Utilized silicon lung casts from canine and mouse models for micro-CT imaging of airway trees.
• Employed a 3D region-growing threshold algorithm (IDL) for airway segmentation from CT data.
• Implemented a fully-parallel 3D thinning algorithm for airway skeletonization and created a tree data structure using Python to store segment lengths.

Main Results:

• Successfully segmented and skeletonized airway trees from micro-CT scans of canine and mouse lung casts.
• Developed a tree data structure capable of storing quantitative geometric information, such as airway segment length.
• Demonstrated accuracy and efficiency for up to six generations in the canine model and ten generations in the mouse model.

Conclusions:

• The developed semi-automatic approach provides an accurate and efficient method for quantitative analysis of small animal airway tree geometry.
• This technique facilitates objective evaluation of bronchial tree structure and function in preclinical research settings.
• The study contributes valuable quantitative morphometric data for small animal respiratory models.