Radiomics Based on Single-Phase CTA for Distinguishing Left Atrial Appendage Thrombus from Circulatory Stasis in Patients with Atrial Fibrillation before Ablation

Figures

• 

  Figure 1. The patient recruitment workflow. CTA, computed tomography angiography; CT, computed tomography; TEE, transesophageal echocardiography; LAAT, left atrial appendage thrombus.

• 

  Figure 2. Radiomics workflow. 3D-VOI, 3-dimensional volume of interest.

• 

  Figure 3. Representative 3D-VOIs segmentation. (A–F) LAA circulatory stasis; (G–L) LAAT; (C,I), non-chicken-wing-shaped LAA; (F,L), chicken-wing-shaped LAA. 3D-VOIs, 3-dimensional volumes of interest; LAA, left atrial appendage; LAAT, left atrial appendage thrombus.

• 

  Figure 4. Comparison of receiver operating characteristic (ROC) curves of all four prediction models for distinguishing left atrial appendage thrombus from circulatory stasis in the training (A) and test (B) set. (A) In the training set, AUC = 0.93 for the Nomo-score; 0.90 for the Rad-radio score. 0.86 for the Rad-clinic score; and 0.82 for the Rad-score; (B) in the test set. AUC = 0.85 for the Nomo-score; 0.84 for the Rad-radio score; 0.82 for the Rad-clinic score; and 0.82 for the Rad-score. ROC, receiver operating characteristic curve; AUC, under of the receiver operating characteristic curve.

• 

  Figure 5. The calibration curves (training set: A; test set: B) and decision curves (training set: C; test set: D) of the combined nomogram. The calibration curves (A,B) depict the agreement of the combined model between the predicted risk and actual probability. The 45° gray line represents the ideal prediction, and the solid black line represents the actual probability of the model. The closer the solid line is to the ideal line, the better the predictive performance of the nomogram is. The decision curve (C,D) is used to assess the clinical utility by calculating the net benefits at different threshold probabilities. The x axis corresponds to the threshold probability while the y axis corresponds to the net benefit. The blue line, red line, green line, and yellow line represents net benefit of the Rad-score, Rad-clinic score, Rad-radio score, and Nomo-score, respectively. The grey line is made with the assumption that all patients are with LAAT. The black line is made with the assumption that no patients are with LAAT. (C) shows that the net benefit of the Nomo-score is better than the other cases when the threshold is not within the range of 78–91% in the training set. (D) Shows that…