Identification of paediatric tuberculosis from airway shape features

Area of Science:

• Medical Imaging
• Pulmonology
• Computer-Aided Diagnosis

Background:

• Paediatric pulmonary tuberculosis (TB) can cause airway stenosis and deformation.
• Accurate detection of airway pathology in children is crucial for effective treatment.

Purpose of the Study:

• To develop and evaluate computer-assisted methods for analyzing airway shape from CT images to detect paediatric TB.
• To assess the efficacy of airway shape analysis in identifying pathological changes.

Main Methods:

• Extraction of airway shape features using principal component analysis (PCA) of the airway surface mesh.
• Analysis of airway branch radius and orientation features.
• Application of support vector machine classification to a dataset of 61 paediatric patients (TB and non-TB).

Main Results:

• PCA-based features (first 10 modes) achieved 86% sensitivity and 91% specificity for TB detection.
• Radius-based features demonstrated 86% sensitivity and 94% specificity.
• Both methods showed high accuracy in differentiating TB from non-TB cases.

Conclusions:

• Computer-assisted airway shape analysis is a potential tool for detecting paediatric pulmonary TB.
• These methods can identify airway pathology resulting from TB and other conditions.
• Further research can refine these techniques for clinical application in diagnosing airway diseases.