Computed Tomography-Based Radiomics Provides New Insights Into Associations Between Pericoronary Fat Characteristics and Low-Density Lipoprotein Cholesterol.

Figures

• 

  Fig. 1.. Automatic drawing of pericoronary adipose tissue (PCAT). LAD, left anterior descending artery; WW, window width; WL, window level; LM, left main; pLAD, proximal Left Anterior Descending; mLAD, mid Left Anterior Descending; dLAD, distal Left Anterior Descending.

• 

  Fig. 2.. Univariate logistic regression results of 11 selected PCAT radiomic features. Abbreviations: CI, confidence interval; FirstOrder, first-order statistics (radiomic feature class); GLSZM, gray level size zone matrix; LCX, left circumflex artery; OR, odds ratio; PCAT, pericoronary adipose tissue; RCA, right coronary artery. Statistical significance: *p < 0.05, **p < 0.01.

• 

  Fig. 3.. Univariate logistic regression forest plot of clinical variables associated with elevated LDL-C levels. Abbreviations: CI, confidence interval; HDL, high-density lipoprotein cholesterol; OR, odds ratio. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001.

• 

  Fig. 4.. Pearson correlation heatmap between PCAT radiomic features and clinical variables. Abbreviations: HDL, high-density lipoprotein cholesterol; PCAT, pericoronary adipose tissue.

• 

  Fig. 5.. ROC curves of the GBM model for training and test sets. Abbreviations: AUC, area under the curve; GBM, gradient boosting machine; ROC, receiver-operating characteristic.

• 

  Fig. 6.. SHAP summary plot of radiomic features associated with elevated LDL-C levels. SHAP values indicate the contribution of each radiomic feature to the discrimination of elevated LDL-C levels, with color representing feature values from low (blue) to high (red). Abbreviations: LDL-C, low-density lipoprotein cholesterol; SHAP, shapley additive explanations.