University of Twente Student Theses


Accelerating MRI with optimized sampling patterns in Pulseq

Schilt, Elise (2024) Accelerating MRI with optimized sampling patterns in Pulseq.

[img] PDF
Abstract:Yearly, about 1 million MRI scans are acquired in the Netherlands [1]. However, a major drawback of using MRI is the long duration of an MRI scan, leading to high costs, waiting lists, and images being affected by motion due to movement of the patient. In MRI, data is acquired in the spatial frequency (k-space) domain. Optimization of k-space sampling patterns, and improved reconstruction methods can reduce scanning time and/or improve image quality. In this work, optimized k-space sampling patterns are investigated, using custom-made pulse sequences in Pulseq [2, 3]. Sampling pattern optimization is done through conventional subsampling of a TSE and radial GRE sequence, as well as using an optimized sampling pattern learned by the deep learning model BJORK [4]. Sequences are successfully executed on a 1.5 T Siemens scanner (Aera, Siemens Healthineers, Erlangen), and data is reconstructed using the BJORK reconstruction model and a NUFFT. The BJORK reconstruction model in this framework shows to be robust to different anatomies, trajectories, and data shapes. Comparison of reconstruction results show the BJORK reconstruction model performs better in reconstructing subsampled data than the NUFFT. The BJORK-optimized sampling pattern was successfully implemented in Pulseq and executed for both phantom and in vivo data. Eddy currents most likely influenced reconstruction results, but this is expected to be mitigated easily. The optimized sampling patterns, both subsampled and BJORK-optimized, showed quick acquisition of data. The acceleration needs to be further investigated by comparing this to state-of-the-art sequences. All in all, this work results in a robust framework to further investigate methods to accelerate MRI.
Item Type:Essay (Master)
Faculty:TNW: Science and Technology
Subject:30 exact sciences in general, 33 physics, 44 medicine, 50 technical science in general, 54 computer science
Programme:Biomedical Engineering MSc (66226)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page