Object: Prediction of aneurysm rupture likelihood is clinically valuable, given that more unruptured aneurysms are being discovered incidentally with the increased use of imaging. The authors set out to evaluate the relative performance of morphological features for rupture status discrimination in the context of the divergent geometrical and hemodynamic characteristics of sidewall- and bifurcation-type aneurysms.
Methods: Catheter 3D rotational angiographic images of 271 consecutive aneurysms (101 ruptured, 135 bifurcation type) were used to assess the following parameters in 3D: maximum diameter (D(max)), height, height/width ratio, aspect ratio, size ratio, nonsphericity index, and inflow angle. Univariate statistics applied to the bifurcation, sidewall, and combined (bifurcation + sidewall) sets identified significant features for inclusion in multivariate analysis yielding area under the curve (AUC) and optimal thresholds in the receiver-operating characteristic. Furthermore, a computational fluid dynamics analysis was performed to evaluate the flow and wall shear stress conditions inside sidewall and bifurcation aneurysms at different inflow angles.
Results: The mean D(max), height, and inflow angle were significantly greater in ruptured sidewall aneurysms than in unruptured sidewall aneurysms, but showed no difference between ruptured and unruptured bifurcation lesions. There was a statistically significant difference between ruptured and unruptured aneurysms for all measured features in the combined set. Multivariate analysis identified the following: 1) nonsphericity index as the only rupture status discriminator in bifurcation lesions (AUC = 0.67); 2) height/width ratio, size ratio, and inflow angle as strong discriminators in sidewall lesions (AUC = 0.87); and 3) height/width ratio, inflow angle, and size ratio as intermediate discriminators in the combined group (AUC = 0.76). Computational fluid dynamics analysis showed that although increasing inflow angle in a sidewall model led to deeper penetration of flow, higher velocities, and higher wall shear stress inside the aneurysm dome, it produced the exact opposite results in a bifurcation model.
Conclusions: Retrospective morphological and hemodynamic analysis point to a dichotomy between sidewall and bifurcation aneurysms with respect to performance of shape and size parameters in identifying rupture status, suggesting the need for aneurysm type-based analyses in future studies. The current most commonly used clinical risk assessment metric, D(max), was found to be of no value in differentiating between ruptured and unruptured bifurcation aneurysms.