Biophysically inspired mechanical biomarkers of normal pressure hydrocephalus - PROJECT SUMMARY While most forms of dementia arise from irreversible, degenerative processes, normal pressure hydrocephalus (NPH) symptoms can often be alleviated by placement of a ventriculoperitoneal shunt to divert cerebrospinal fluid from the characteristically enlarged ventricles. However, due to overlapping clinical and biomarker phenotypes with other common age-related disorders, NPH remains substantially underdiagnosed and undertreated. Even when diagnosed, current methods are limited in predicting outcomes of shunt surgery, particularly when considering noninvasive methods. Since the disorder was first described, it has been hypothesized that the intracranial mechanical environment was involved in NPH pathogenesis. Consistent with this hypothesis, we recently reported that NPH is associated with a characteristic pattern of brain viscoelasticity, as measured by magnetic resonance elastography (MRE). Furthermore, the presence of this pattern is sensitive and specific for differentiating NPH patients from healthy volunteers and those with Alzheimer’s clinical syndrome. While MRE is an accurate diagnostic biomarker for NPH, we expect that with further technical development, MRE will also enable accurate prediction of treatment outcomes. In our preliminary data, patients with low brain stiffness are less likely to benefit from shunting. However, these predictions are not yet accurate enough to use clinically. We hypothesize that the existing discrepancies arise from an ambiguity that low brain stiffness, as measured by current technology, can result either from decreased stiffness in the solid tissue matrix or increased fluid content in the extracellular space. These two effects must be disentangled to identify the patients with preserved matrix stiffness as those most likely to benefit from shunting. Therefore, the overall goal of this work is to develop an MRE framework that jointly leverages strain and diffusion measurements to estimate the viscoelastic properties of the solid matrix while accounting for interspersed extracellular fluid. In Aim 1, we will modify the forward model of our machine learning-based inversion framework to account for subvoxel fluid elements. We will then evaluate the algorithm’s accuracy in simulation and phantom experiments, along with its repeatability in vivo. In Aim 2, we will compare the diagnostic accuracy of the new method against existing methods, confirming that this solid-fluid mixture framework can similarly discriminate patients with NPH from healthy controls and those with Alzheimer’s dementia. Finally, using objective measures of gait collected before and after surgery, we will test the overall study hypothesis that baseline solid matrix stiffness estimates (accounting for fluid) can predict the degree of gait improvement following shunt placement. The proposed technology represents a fundamental shift in the field of brain MRE toward biophysically inspired modeling. Most importantly, the success of this proposal will provide new insights into the pathogenesis of NPH with the potential to directly impact patient care.