Computer-aided detection and quantification of cavitary tuberculosis from CT scans.

Area of Science:

• Medical Imaging
• Radiology
• Computer-Aided Diagnosis

Background:

• Tuberculosis (TB) cavities in the lungs are critical indicators of disease severity and transmission.
• Accurate identification and quantification of these cavities from CT scans are essential for patient management.
• Existing methods may lack the precision or automation required for comprehensive analysis.

Purpose of the Study:

• To develop and evaluate a computer-aided detection (CAD) tool for identifying, quantifying, and assessing TB cavities in lung CT scans.
• To analyze the spatial relationship between cavities and airways and track longitudinal changes in cavity morphology.
• To provide a fully automated system for TB cavity analysis.

Main Methods:

• A novel shape-based automated detection algorithm combined with fuzzy connectedness (FC) for cavity and airway delineation.
• Utilized support vector machine classification for structure detection and a hybrid multiscale approach for airway tree extraction.
• Employed intensity-based FC for cavity segmentation and linear time distance transform for spatial analysis.

Main Results:

• The CAD tool achieved high accuracy in cavity detection (94.61%) and airway detection (99.8%).
• Excellent agreement was observed between automated segmentation and expert radiologists (Dice similarity ~99.0%, volume correlation R² > 0.99).
• The system precisely quantified cavity volume, surface area, and spatial relationships with airways, enabling longitudinal analysis of morphological evolution.

Conclusions:

• A fully automated method for TB cavity detection, quantification, and evaluation using CT scans has been successfully developed.
• The proposed framework offers high accuracy, efficiency, and provides novel insights into cavity formation and its relationship with airways.
• This represents a significant advancement in the computerized analysis of cavitary tuberculosis.