Texture Analysis Contribution to Evaluate the Common Carotid Artery’s Cardiovascular Disease (CVD) Risk Using Structural Equation Modeling

Clinical cardiovascular disease (CVD), which may increase the risk of stroke may be evaluated using the common carotid artery’s (CCA), the intima media thickness (IMT) and textural characteristics extracted from the CCA’s intima media complex (IMC, the artery wall). Using structural equation modeling (SEM), this study analyzes the relationship between the IMT and textural features of the IMC of the CCA and the prevalent clinical CVD. The study used 612 longitudinal-section ultrasound images of the left and right CCA from 158 men and 148 women, 42 of whom had clinical CVD. Images were intensity normalized and despeckled. For all images, the IMC was semi-automatically segmented using an in-house semi-automated segmentation system, and 40 different texture features were retrieved. To that purpose, we suggested a novel method for analyzing the above features and calculating the CVD risk. In this investigation, SEM was used to create a theoretical model of correlations between eight different elements (unobserved constructs and observable feature variables). More specifically, six different IMC texture feature groups, derived from the IMC of the CCA in ultrasound images, as well as the IMT, and the CVD were taken into consideration. The primary conclusions of the study are as follows: (i) The six IMC texture feature groups (factors) tested in conjunction with IMT fit the conceptual model very well. (ii) The conceptual model’s seven hypothesized paths for the impact of each texture feature group on CVD were tested. Six of the selected factors were shown to have a substantial impact on CVD, two of which with p < 0.05 (Spatial Gray Level Dependence (SGLDM), IMT) and four with p < 0.10 (90% confidence level). The findings of this study significantly improved upon those previously reported because of the very good model fit (e.g., normed fit index (NFI) = 0.94). They might provide further complementary data for CVD risk modelling.