Solitary Pulmonary Nodule Characterization on CT by use of Contrast Enhancement Maps

摘要

Studies have shown that vascular structure of a solitary pulmonary nodule (SPN) can give insight into the diagnosis of the nodule. The purpose of this study is to investigate the utility of texture analysis as a quantitative measure of the vascular structure of a nodule. A contrast CT study was conducted for 29 patients with an indeterminate SPN. For each patient, the post-contrast series at maximum enhancement was volumetrically registered to the pre-contrast series. The two registered series were subtracted to form difference images of the nodule and each voxel was color-coded into 7 bins. Initially, a representative image of each nodule was subjectively rated on a five-point by a radiologist as to the magnitude, extent, and heterogeneity of the enhancement. From the initial analysis the heterogeneity of the nodule was found to be significantly different for benign versus malignant nodules (p<0.01), while the other two ratings were found not to be significant. We then attempted to quantify this subjective rating of heterogeneity by calculating 14 textural features based on co-occurrence matrices. These features included various measures of contrast, entropy, energy, etc. Dimension reduction techniques such as principal component and factor analysis were applied to the features to reduce the 14 variables to one factor. The mean of this factor was significantly different for malignant versus benign nodules (p=0.010). Texture analysis of contrast enhancement maps appears to be useful tool to characterize SPNs.