Magnetic resonance elastography (MRE) is a new method for assessing the properties of the brain by imaging of its mechanical properties, such as stiffness and its ability to dissipate energy. By observing how the brain vibrates due to an external driver, MRE provides rich information about the tissue that can be used to infer information regarding the brain's microscopic composition which is not accessible by other means using in-vivo methods. To obtain an MRE measurement the head of a subject is vibrated inside of an MR scanner with the use of a vibrating pillow while the phase of the signal is monitored, from which tissue displacement is obtained over multiple time-points creating a propagating wave image inside the brain. Currently, we are the only group in Sweden with the technical capabilities of performing these MRE scans of the brain, and we are thus involved in several clinical and technical projects. Clinical projects include, but not limited to, investigating the mechanical and microstructural changes due to neurodegenerative diseases such as Parkinson's disease, Alzheimer's disease, or multiple sclerosis. For technical development we are working on several methods attempting to either improve performance of current MRE technologies, or develop models that extract additional information from the MRE images. One of the most challenging aspects of MRE is obtaining accurate mechanical properties from the acquired wave images. This is typically done using various inversion algorithms with pros and cons. We routinely use a form of fast inversion that relies on several assumptions of the material, such as local homogeneity and that the material is isotropic. However, these assumptions can lead to significant errors, thus other methods have been developed. One such method, which is highly flexible and relies on much fewer assumptions, is the non-linear inversion (NLI) method developed by McGarry et al (Medical Physics, 2012). This method creates a basic guess of the material properties of the brain and then calculates the resulting wave pattern in the brain using a forward solve and then compares that solution to experimental data, and then updates the material properties according to the difference between real and simulated data. This is done iteratively until a sufficiently accurate map of the material properties has been obtained. Although computationally expensive, this inversion method provides the most reliable results. These are needed to provide accurate assessments to our clinical collaborators, but also to be used as benchmarks for further method development. One example of such method development is to use in-vivo data to train a convolutional neural network to perform the inversion in a fraction of the time NLI does, however to do so we need to have better ground truth, which NLI can provide. Several different projects would benefit greatly from having NLI estimates of the mechanical properties of the brain from MRE data, and in order to obtain this we require additional computational resources.