We investigated the performance of CNN-assisted joint estimation in two cases of severe motion corruption in a 2D slice of T2w FSE MRI. We showed that the inclusion of the CNN can help speed up convergence of the joint estimation algorithm, corroborating previous findings. We also showed one case in which joint estimation failed to converge to the correct image and motion parameters, with and without the CNN. A more exhaustive study is required to confirm whether deep learning can help joint estimation salvage otherwise unsalvageable corrupted data.
This abstract and the presentation materials are available to members only; a login is required.