Statistical textural features for classification of lung emphysema in CT images: A comparative study

Computed tomography (CT) can contribute to the early detection of lung diseases like emphysema, a chronic and progressive disease. Texture-based methods can be explored to classify regions of interest (ROI´s) into emphysematous areas and normal areas. In this work we evaluated the importance of a set of parameters in the classification of lung CT images, such as the size of the ROIs, the quantization level, and textural features used in classification. A support vector machine was used as classifier. The performance of the designed classifier was evaluated using a 10-fold cross validation method and the results compared based on overall accuracy, sensibility and specificity. This study shows that textural features have a good discriminatory power in the classification of lung emphysema in CT images.