Imaging the microstructure of brain gray matter

MRI combined with biophysical modeling enables 3D noninvasive mapping of brain tissue microstructure, i.e., quantitative imaging of tissue properties on a scale of micrometres, orders of magnitude below the nominal image resolution of millimeters. The ability to do so crucially depends on adequate and validated biophysical modeling, and robust analysis of multiple MRI images. Diffusion weighted MRI (dMRI) is a particularly well-suited type of MRI because the water molecules probe their microenvironment on a length scale on the order of ~10 µm during signal encoding. Thus, the signal indirectly reflects details of tissue structure on that length scale, which combined with modeling enables mapping of tissue microstructure over the whole brain – “MRI Virtual Microscopy”, MVM.

The Standard Model of diffusion in white matter has been particularly successful for MVM, mapping intra/extra-axonal diffusivities, axon density, and axon orientation distribution – all properties which are normally accessible only from looking at tissue biopsies through a microscope.

However, in gray matter, our understanding of the interplay between microstructure and MRI signals is less developed. This is presumably due to more complex and heterogenous microstructure, as well as substantial water diffusion across neurite membranes. Some progress with new models has been made recently in our group and elsewhere, but they remain simplified and have not yet been validated. Nevertheless, MVM of gray matter could have significant impact for early diagnosis and understanding of diseases involving cortical tissue or deep gray matter nuclei, such as ALS and Parkinson.

Therefore, the aim of the MSCA would be to advance MVM of gray matter. This could for example be done by :

• Validating existing models:
  • with ultrarealistic computer simulations based on histology
  • with experiments utilizing tissue perturbations such as membrane pump inhibitors
  • comparison to histology  
• Model estimation procedures with machine learning
• Adapting and optimizing protocols with generalized q-space encoding
• Exploring human translation and potential application in disease cohorts, e.g. ALS
• Proposing and testing new models, especially incorporating time dependence due to structure as well as exchange
• Comparing predictions to other types of (d)MRI such as Correlation Tensor Imaging, Oscillating Gradients etc.

MRI and some of the following:

• Animal MRI
• Human MRI
• dMRI
• Computer simulation (monte carlo, finite difference/elements etc.)
• Machine learning
• MRI data analysis
• Modeling
• Histology

• Microstructure MRI
• Diffusion
• Susceptibility
• MVM
• Biophysical modeling
• Computer simulations
• MRI
• Brain microstructure

More about the host group