Development of an automatic classification system for differentiation of obstructive lung disease using HRCT

This study aims to improve the performance of differentiating obstructive lung diseases based on high-resolution computerized tomography (HRCT) images by introducing new shape features and optimizing the classifier. The study utilized 265 HRCT images from 82 subjects, and two experienced radiologists selected regions of interest (ROIs) representing areas of severe centrilobular emphysema, mild centrilobular emphysema, bronchiolitis obliterans, or normal lung. In addition to 13 textural features, 11 shape features were employed to evaluate their contribution. The Bayesian classifier and support vector machine (SVM) were implemented, and a five-folding method was used to assess cross-validation of the system. The study found that adding shape features to conventional texture features significantly improved the overall sensitivity of both the Bayesian and SVM classifiers, particularly in smaller ROIs. The results suggest that shape features contribute more to overall sensitivity in smaller ROI sizes. The findings of this study have implications for the development of AVIEW Lung Texture, as it highlights the usefulness of adding shape features to conventional texture features for improving classification performance of obstructive lung diseases.