The consequence of fiber orientation downsampling on the computation of white matter tract-related deformation
Incorporating neuroimaging-revealed structural details into finite element (FE) head models opens vast opportunities to understand brain injury mechanisms. Recently, growing efforts have been made to integrate the fiber orientation from diffusion tensor imaging into the FE models to compute white matter (WM) tract-related deformation. Commonly used approaches often downsample the spatially enriched fiber orientation to match the resolution of FE meshes, resulting in an element-wise orientation implementation. However, the validity of downsampling and the consequences on the computed tract-related strains remain elusive. To address this problem, the current study proposed a new voxel-wise approach to integrate fiber orientation into FE models without downsampling. By setting the voxel-wise orientation responses as the reference, we then evaluated the reliability of two existing downsampling approaches on tract-related strains using two FE models with varying element sizes. The results showed that, for a model with a large mesh-image resolution dismatch, the downsampling orientation exhibited an absolute difference over 30 degree across the WM/gray matter interface and pons regions and further negatively affects the computation of tract-related strains with the normalized root-mean-square error up to 20% and peaking tract-related strains underestimated by 5%. This downsampling-induced effect was lower in FE models with finer meshes. Thus, this study yields insights on integrating neuroimaging-revealed fiber orientation into FE models and may better inform the computation of WM tract-related deformation, which are crucial for advancing the etiological understanding and computational predictability of brain injury.