Global, whole-brain model validation

Revising the material properties of the brain required fresh model validations. The previous WHIM with an isotropic Ogden material model was validated using three cadaveric impacts along three different directions (frontal (C383-T1), occipital (C755-T2), and parietal (C393-T4) impacts;^11,46). The evaluations were limited to corner locations of neutral density targets (NDT) because of data availability in the original publications.^61,62 As more NDT displacement data have become available in other studies, here we reported WHIM validation performances using an expanded dataset (additional tests included C383-T3, C383-T4, and C291-T1; 6 tests in total) for all of the available NDT locations (N = 56 in total).⁴³ For each selected cadaveric impact, WHIM was scaled to match the reported head size. The resulting model-predicted brain-skull relative displacement time histories were then compared against those measured in experiments.

In addition to high-rate cadaveric impacts, strain data from an in vivo head rotation⁴⁴ was also used to validate at the opposite extreme of impact severity. Radial-circumferential shear strains were obtained from the simulation to report the above-threshold area fractions. For all simulations, validation performances were assessed using correlation score (CS) for consistency with the previous work¹¹ as well as CORrelation and Analysis (CORA⁶⁵). CS and CORA scores were reported for both the previous isotropic WHIM (Ogden) and the current anisotropic version (HGO), using either DTI voxels or tractography for WM anisotropy implementation.

Further, we used CORA as a ranking metric⁴³ to compare WHIM validation performances against other recent or well-utilized models, including ABM,¹⁷ SIMon,²¹ THUMS,⁶⁶ GHBMC,¹⁶ and the KTH model (with two isotropic material models and an additional anisotropic model;^19,29).

Finally, validation against impact-induced brain pressure responses was not necessary for a nearly incompressible brain,¹ because the pressure responses were essentially hydrostatic for blunt impacts with a duration longer than 2 msec.^67,68 They can be determined uniquely by the magnitude and directionality of linear acceleration, as well as brain size and shape for a given head model.^49,69