University of Twente Student Theses


Joint reconstruction and motion estimation via nonconvex optimization for dynamic MRI

Bastiaansen, W.A.P. (2018) Joint reconstruction and motion estimation via nonconvex optimization for dynamic MRI.

[img] PDF
Abstract:Dynamic magnetic resonance imaging (MRI) is a medical imaging technique. MRI reconstruction methods build upon inverse problems, variational methods and optimization in applied mathematcis. To reduce scanning time in dynamic MRI, subsampling of the measurements is needed in practice. This typically leads to artefacts due to missing information. To tackle those artefacts, time-dependent reconstruction methods, which employ not only the spatial properties of the image sequence, but also the dynamical information, are very promising. In this thesis a joint reconstruction and motion estimation framework is developed and applied to dynamic MRI reconstruction. The optimization of this joint variational model is challenging since it is nonconvex. Current approaches alternate between the convex subproblems for reconstruction and motion estimation. This approach seems to be working but there is no knowledge about the convergence. To address this problem the full nonconvex model is optimized via an alternating forward-backward splitting algorithm which is related to the PALM algorithm for nonconvex optimization. The performance of joint reconstruction and motion estimation on dynamic MRI is studied. In real medical datasets no ground truth of the flow fields is available. To address this, artificial data from computer vision, with a ground truth for the flow fields at hand, is used to construct a numerical phantom. In combination with a 4D XCAT phantom, this offers more insight into the influence of incorporating the motion on the reconstruction quality and choice of optimization method. Finally, the model is applied to experimental medical data from the Radiotherapy group of the UMC Utrecht to show its potential for real world scenarios.
Item Type:Essay (Master)
Faculty:EEMCS: Electrical Engineering, Mathematics and Computer Science
Subject:31 mathematics
Programme:Applied Mathematics MSc (60348)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page