Prediction of Thrombosis After Coil Embolization of Unruptured Cerebral Aneurysm by Computational Fluid Dynamics Using Porous Media Modeling

Abstract

Purpose

Preoperative simulations tailored to the therapeutic device are desirable to predict treatment efficacy. We investigated the hemodynamics associated with complete obliteration (CO) of intracranial aneurysms after endovascular treatment by computational fluid dynamics (CFD) using porous media modeling.

Methods

From June 2015 to December 2020, 48 patients with 55 unruptured aneurysms treated with simple coiling or stent-assisted coiling at our institution were included. Patient-specific geometry models were generated from pre-treatment 3 dimension (D) rotational angiography or 3D computed tomography angiography. We simulated the coil and neck-bridging stent with porous media modeling for steady state analysis and calculated residual flow volume (RFV) associated with thrombosis formation after coil embolization. Patients were classified into CO and non-CO groups based on digital subtraction angiography findings after 6 months of treatment, and the groups were statistically evaluated using the Brunner-Munzel test. The area under the receiver operating characteristic (ROC) curve (AUC) was used to evaluate the diagnostic accuracy in predicting CO.

Results

The aneurysms were classified into 22 CO and 33 non-CO group. There were no significant differences in morphological variables between the two groups. In hemodynamic parameters, RFV was significantly lower in the CO group. The ROC curve analyses showed that RFV with an average flow velocity of more than 0.5 [cm/s] in aneurysm was the most useful to predict CO (AUC, 0.66 [95% CI, 0.51–0.81]; cut-off value, 32.8 mm³; sensitivity, 60.6%; and specificity, 78.9%).

Conclusion

These results indicated that CFD could predict aneurysmal CO after coil embolization.