Meeting Banner
Abstract #4471

Semi-Joint Reconstruction for Diffusion MRI Denoising Imposing Similarity of Edges in Similar Diffusion-Weighted Images

Vladimir Golkov 1,2 , Marion I. Menzel 1 , Tim Sprenger 1,3 , Axel Haase 3 , Daniel Cremers 2 , and Jonathan I. Sperl 1

1 Diagnostics & Biomedical Technologies - Europe, GE Global Research, Garching n. Munich, Germany, 2 Department of Computer Science, Technische Universitt Mnchen, Garching n. Munich, Germany, 3 Institute of Medical Engineering, Technische Universitt Mnchen, Garching n. Munich, Germany

Recently, Joint Reconstruction has been proposed by Haldar et al. for SNR enhancement of diffusion-weighted images (DWIs), performing edge-preserving denoising by imposing identical edge constraints to all DWIs. In this work, we propose Semi-Joint Reconstruction to allow individual edges for each DWI in order to account for the fact that distinct DWIs can look quite dissimilarly, whereas similar DWIs contain common edge structures. Individual edge maps necessitate edge map denoising, for which we use DWI similarity weightings, truncated singular value decomposition of DWIs, and shearlet-based edge detectors. Results are comparable to Joint Reconstruction, but more stable to regularization parameter choice.

This abstract and the presentation materials are available to members only; a login is required.

Join Here