Computer aided diagnosis of small pulmonary nodules based on hierarchical classification using contrast enhanced dynamic CT images

摘要

This paper presents a computerized classification scheme of pulmonary nodules in contrast enhanced dynamic CT images. In this scheme, we introduce a clustering procedure of 3-D nodule images as a preprocess of discriminating between malignant and benign nodules. The feature vector used in the clustering procedure consists of the mean CT density values in two different regions of the 3-D nodule and the nodule volume. The 3-D nodule images in each cluster is discriminated between malignant and benign nodules by using the curvature based features. The potential usefulness of the scheme using the contrast enhance dynamic CT images is demonstrated by using discriminate the score plotting and receiver operating characteristic (ROC) curves as the performance measure.