We propose a deep learning model that is able to separate a tractogram into sets of anatomically plausible and implausible streamlines. In contrast to existing methods, our model relies solely on the measured diffusion signal as an input ensuring independence of potential misalignments between subjects. The model is shown to generalize to different tractography methods and has the potential to simultaneously learn from multiple supervisor methods.
This abstract and the presentation materials are available to members only; a login is required.