Hybrid segmentation and virtual bronchoscopy based on CT images.

Area of Science:

• Medical Imaging
• Pulmonology
• Computer-Aided Diagnosis

Background:

• Virtual endoscopy offers a non-invasive method for visualizing internal anatomical structures.
• Accurate segmentation of the tracheobronchial tree is crucial for effective virtual endoscopic analysis.
• Existing methods may have limitations in visualizing complex or hidden pulmonary structures.

Purpose of the Study:

• To introduce and evaluate a novel virtual bronchoscopy system integrating SegoMeTex segmentation and VIVENDI.
• To enable detailed visualization of the tracheobronchial tree down to the seventh generation.
• To demonstrate the system's capability in visualizing hidden structures like vascular systems and tumors.

Main Methods:

• Utilized multislice computed tomography (CT) data for segmentation.
• Employed a hybrid SegoMeTex system for automated tracheobronchial tree segmentation with minimal user input.
• Manually marked additional structures such as tumors.
• Generated surface data structures for virtual endoscopy.
• Explored datasets using the VIVENDI virtual bronchoscopy system.

Main Results:

• Successfully tested on 22 patients.
• Achieved high sensitivity (>58%) and positive predictive value (>90%) for identifying bronchi up to the sixth generation.
• Demonstrated real-time interactive exploration (>30 fps) on standard PCs.
• Showcased high-quality reconstruction of even small anatomical structures.

Conclusions:

• The combined virtual bronchoscopy and segmentation system is a valuable tool.
• It facilitates accurate localization and measurement of stenosis.
• This technology supports effective resection planning for lung conditions.