Unfortunately, the state-of-the-art "in vivo" imaging technology with its millimeter-scale resolution lags behind by 1-2 orders of magnitude to reveal the cortical layers directly. However, there is evidence that signals corresponding to distinct layers can be distinguished in individual voxels with numerical analysis even in relatively low-resolution MRI data of a specific measurement protocol. The analysis yields probabilities of occurrence of all the layers in each voxel, which can be used to reconstruct the layers at subvoxel level.

The student will develop a robust implementation of this numerical analysis, taking the raw MRI data as input and producing 3D array of layer probabilities in a format suitable for further visualization and layer reconstruction. The thesis will discuss the correctness of the approach, and it will design an appropriate testing protocol to support it.