Charles M. Strother

Over the last several years the number of reports related to computational fluid dynamics (CFD) in the AJNR has increased significantly. A majority concerned intracranial aneurysms (IAs), with the primary focus being on development and validations of a technique that could allow clinically useful predictions of IA risk of rupture. Although several papers on this topic have shown the ability to distinguish aneurysms that have ruptured from ones that have not, to my knowledge, none, neither in the AJNR nor in other literature, have shown utility in predicting the natural history of IAs, ie, rupture risk. While the goal of using CFD to identify a parameter or combination of parameters predictive of rupture risk remains elusive, increasing knowledge of the material properties of the aneurysm wall, increased understanding of the links between blood flow and arterial homeostasis/vascular remodeling, and added sophistication to computational simulations all serve as motivation to continue with and validate these techniques.

In this issue of the AJNR Digest, papers by Valen-Sendstad and Steinman and by Jansen and colleagues describe and illustrate how the sensitivity and reliability of computational studies can be, to a significant degree, dependent upon using proper solution strategies and boundary conditions. An earlier paper from the group in Amsterdam (Schneiders et al) addressed the fundamental issue of geometry correctness when using rotational DSA data for the construction of vascular models. The paper by Byrne and

colleagues is an excellent example, in a large series of IAs, of using computational techniques to search for newly demarcated hemodynamic characteristics (in this case, spatial complexity and temporal stability) that might be useful in prediction of rupture risk. In the final paper in this series, Geers and colleagues found that, for the techniques used in their simulations, the use of vascular constructions based on multidetector CTA and rotational DSA yielded large differences in values for all hemodynamic parameters that were evaluated quantitatively.

Because signals from hemodynamic forces are translated into the biologic responses that direct arterial homeostasis/vascular remodeling, there is reason to believe within these signals are ones that, perhaps, might be indicative of why some aneurysms rupture but most do not. The greater the “realism” used to carry out computational studies the greater the likelihood becomes that the results will truly reflect the actual forces to which IAs are exposed. As more is learned about the material properties of the aneurysm wall, this progress should allow performance of more “patient-specific” fluid structure interaction simulations. There is reason to believe that within the resulting output of hemodynamic and biomechanical parameters combinations exist that may have added utility in determining rupture risk.

Today, computational studies have limited availability, as well as proven usefulness in the clinic. As the ability to perform simulations becomes available on commercial platforms in timeframes that have clinical relevance, availability will increase. Finding and proving clinical utility for both these new “clinical” applications as well as for “conventional” methods depends upon the results of carefully designed trials using appropriate computational techniques and boundary conditions. There is a real danger that, enamored by beautiful images and quantitative values, those not fully aware of the complexities and limitations inherent in CFD simulations may try to extract more information than is truly there. I hope that this issue of the AJNR Digest will serve to emphasize that in CFD studies, detail matters a great deal.

Image modified from: Schneiders et al. Intracranial aneurysm neck size overestimation with 3D rotational angiography: the impact on intra-aneurysmal hemodynamics simulated with computational fluid dynamics.