The role of epicardial and pericoronary adipose tissue radiomics in identifying patients with non-ST-segment elevation myocardial infarction from unstable angina

Figures

• 

  Fig. 1. A flowchart of patient selection and study design. All patients were separated into training cohort (n = 116; Fig. 1A), internal validation cohort 1 (n = 50; Fig. 1A), and internal validation cohort 2 (n = 50; Fig. 1B) based on the time order of admission. The internal validation cohort 1 and the training cohort used the same scanner and scan parameters, while the internal validation cohort 2 used different canners and scan parameters than the training cohort. ACS, acute coronary syndrome; CCTA, coronary computed tomography angiography; ICA, invasive coronary angiography; NSTEMI, non-ST-segment elevation myocardial infarction; UA, unstable angina; BMI, body mass index.

• 

  Fig. 2. The radiomics workflow of this study. (A) The ROIs of EAT and PCAT were automatically segmented and then manually adjusted appropriately. (B) Radiomics features were extracted, including first-order, shape, and texture features (GLCM, GLRLM, GLSZM, GLDM, NGTDM). (C) ICC, mRMR, and LASSO were used for feature selection. Next, a logistic regression model was constructed using the selected features. (D) All models were evaluated using the ROC curve, calibration curve, and DCA. EAT, epicardial adipose tissue; PCAT, pericoronary adipose tissue; ROI, region of interest; RCA, right coronary artery; LAD, left anterior descending artery; LCX, left circumflex artery; First-order, first-order features; Shape, shape features; GLCM, gray level cooccurrence matrix; GLRLM, gray level run length matrix; GLSZM, gray level size zone matrix; GLDM, gray level dependence matrix; NGTDM, neighboring gray-tone difference matrix; ICC, intraclass correlation coefficient; mRMR, maximum relevance minimum redundancy; LASSO, least absolute shrinkage and selection operator; ROC, receiver operating characteristic; DCA, decision curve analysis.

• 

  Fig. 3. The coefficient weight map of the combined model. With the rad-scores of the RCA, LAD, and LCX models, the combined model was constructed in the training cohort. RCA, right coronary artery; LAD, left anterior descending artery; LCX, left circumflex artery; rad-score, radiomics score.

• 

  Fig. 4. The ROC curves of all models in the training cohort (A), internal validation cohort 1 (B), and internal validation cohort 2 (C). ROC, receiver operating characteristic; EAT, epicardial adipose tissue; RCA, right coronary artery; LAD, left anterior descending artery; LCX, left circumflex artery; AUC, area under the receiver operating characteristic curve.

• 

  Fig. 5. Delong test for AUCs of all models in the training cohort (A), internal validation cohort 1 (B), and internal validation cohort 2 (C). AUC, area under the receiver operating characteristic curve; EAT, epicardial adipose tissue; PCAT, pericoronary adipose tissue; RCA, right coronary artery; LAD, left anterior descending artery; LCX, left circumflex artery.

• 

  Fig. 6. The calibration curves of all models in the training cohort (A), internal validation cohort 1 (B), and internal validation cohort 2 (C). The Brier score was computed to assess the overall performance of all models. The combined model had the lowest Brier score of all the models, and the RCA-PCAT model had a lower Brier score than the EAT model, LAD-PCAT model, and LCX-PCAT model. EAT, epicardial adipose tissue; PCAT, pericoronary adipose tissue; RCA, right coronary artery; LAD, left anterior descending artery; LCX, left circumflex artery; Brier, Brier score.

• 

  Fig. 7. The decision curve analysis of all models in training cohort (A), internal validation cohort 1 (B), and internal validation cohort 2 (C). The DCA curves showed that the combined model distinguished NSTEMI from UA with a better overall net benefit than other models. EAT, epicardial adipose tissue; PCAT, pericoronary adipose tissue; RCA, right coronary artery; LAD, left anterior descending artery; LCX, left circumflex artery; DCA, decision curve analysis.