Image‐based modeling of biomechanical factors for risk assessment of developing periventricular white matter hyperintensities

Abstract

Abstract Background White matter lesions, visible as white matter hyperintensities (WMH) on T2‐weighted MR images, have been associated with aging and with cognitive decline. Among WMHs, periventricular WMHs (adjacent to the ventricular system) have preferential associations with cognitive decline/impairment. Despite the prevalence and potential significance in cognitive declines, there is little to no in‐depth knowledge regarding the underlying causes of WMH, except that it is related to small vessel disease. We hypothesize that the development of periventricular WMH is caused by pulsatile stresses exerted on the lateral wall of the ventricles. Elevated stress in regions adjacent to the ventricular wall, coupled with the weakening of the ependymal layer, may trigger the early changes in white‐matter microstructure and surrounding small vessels, eventually causing WMH. Method We developed a method to predict stress distribution in the periventricular white matter using patient‐specific brain anatomy and ventricular movement, obtained through MR imaging techniques. Utilizing established tools for image‐based segmentation (ITK Snap), pre‐processing (3D Systems–Geomagic; Altair–Hypermesh), and finite element modeling (ANSYS Workbench), we tested this pipeline on a healthy volunteer. Patient‐specific ventricular wall movement was assessed over the cardiac cycle using fast EPI MRI and utilized as a boundary condition driving displacement on the ventricular walls in the computational model. Result As shown in Figure 1, the stress distribution around the lateral ventricles was calculated based on subject specific ventricular geometry. The stress distribution is not uniform, but changes with the curvature of the ventricular wall. Peak stress concentrations were detected around the anterior and posterior horns of the lateral ventricles, where WMHs are commonly found. Initial results provide support for our hypothesis of biomechanical contributions to lesion development. Conclusion We developed a computational modeling framework based on the ventricular anatomy obtained with structural MR images, and material properties from relevant literature and fast EPI MRI. The periventricular locations with high stresses were correlated with the common WMH topography. In the next step, we will apply the model to patient data to further determine the role of biomechanical factors in forming WMH.