Objective: The Watchman 2.5 occluder device is a useful device to treat atrial fibrillation (AF), and predicting the size of the Watchman 2.5 occluder device is important to the therapeutic efficacy. To use cardiac computed tomography angiography (CCTA) to predict the size of a Watchman 2.5 occluder device is a potential approach.
Methods: The CCTA was used to individually plan the left atrial appendage (LAA) landing zone and measure the longest and shortest diameters, in addition to the perimeter. The average diameter, the perimeter-derived diameter (PDD), and the ellipticity index (EI) are then calculated from the above values. The longest diameter, the shortest diameter, the average diameter, and PDD of the landing zone were used to predict the occluder size. The size of the occluder was predicted using the longest diameter, the shortest diameter, the average diameter, and the PDD, which is then compared to the actual size.
Results: There were differences between the predicted and actual values of the four groups, with the smallest variability in PDD (P = .007). There was a strong positive correlation between the four groups and the actual occluder size, with the strongest PDD correlation (r = 0.941, P < .001). The prediction accuracy ranged from 44.1% to 90.1% for different methods, with PDD having the highest prediction accuracy. The ROC curve of EI was predicted and plotted using the longest diameter method recommended in the Watchman's instructions, and the area under the curve was 0.905 (95%Confidence Interval (CI) 0.847-0.963), with a cut-off value of 1.198, a sensitivity of 88.9% and a specificity of 82.7%. LAAs with an EI<1.198 had similar accuracy in predicting occluder size, regardless of whether the longest diameter (93.94%) or PDD (87.88%) (P = .344) method was used. However, the kappa test showed poor agreement between the two methods (Kappa = 0.093). When EI ≥ 1.198 (n = 45), the accuracy of PDD in predicting occluder size was 93.33%, which is significantly higher than predictions of the longest diameter (28.9%) (P < .001).
Conclusions: The longest diameter and the PDD methods predicted occluder size with a high degree of accuracy when the LAAs EI < 1.198; the PDD method is suggested to be a preferred method to treatAF.