An Automated Broncho-Arterial (BA) Pair Segmentation Process and Assessment of BA Ratios in Children with...

Area of Science:

• Medical Imaging
• Pulmonology
• Artificial Intelligence in Medicine

Background:

• Bronchiectasis in children requires early diagnosis and treatment to prevent severe lung disease.
• The broncho-arterial (BA) ratio derived from high-resolution computed tomography (HRCT) scans is a key radiological diagnostic criterion for bronchiectasis.
• Accurate detection of broncho-arterial pairs is crucial for calculating the BA ratio.

Purpose of the Study:

• To develop an automated approach for identifying potential broncho-arterial (BA) pairs from pediatric HRCT scans.
• To facilitate the diagnosis of bronchiectasis through efficient BA pair detection and BA ratio calculation.
• To evaluate the robustness of the automated BA detection method using a deep learning model.

Main Methods:

• Lung segmentation and histogram analysis-based cleaning of HRCT scans.
• Identification of potential arteries based on specific imaging features and extraction of connected components.
• Identification of potential bronchi and matching with arteries to form BA pairs.
• Calculation of BA ratios using measured diameters and areas of segmented bronchi and arteries.

Main Results:

• An automated method successfully detected 8-50 BA pairs per patient from HRCT datasets.
• A deep learning model achieved a high classification test accuracy of 98.53%, validating the automated BA detection approach.
• The automated segmentation and BA ratio calculation show potential for aiding in the diagnosis of pediatric bronchiectasis.

Conclusions:

• Visible broncho-arterial pairs can be automatically identified and segmented from pediatric HRCT scans.
• The automated calculation of the broncho-arterial ratio offers a promising tool for the diagnosis of bronchiectasis.
• This automated approach can potentially reduce diagnostic effort and time for pediatric bronchiectasis.