Advanced Motion Corrected Reconstruction Techniques for Magnetic Resonance Imaging

Student thesis: Doctoral ThesisDoctor of Philosophy


Magnetic Resonance Imaging (MRI) is a powerful imaging modality with excellent soft tissue contrast and high spatial resolution without the need for ionising radiation. However, the acquisition process is inherently slow, which imposes practical constraints on the modality. Scan times are particularly long in three-dimensional high spatial resolution imaging. This diculty has recently been alleviated by accelerated acquisitions, combined with Parallel Imaging or Compressed Sensing reconstructions.

Patient motion is one of the major obstacles in clinical MRI, as physiological motion is typically faster than the acquisition process. Motion occurring during a scan will corrupt the acquired data and introduce image artefacts in the reconstructed image. Unavoidable types of motion such as respiratory motion must be considered for a successful MR examination. The problem of respiratory motion is most predominant in abdominal and cardiac imaging. To tackle this concern, motion corrupted data is commonly rejected using the so-called gated data acquisition. However, scan times are increased further as rejected data needs to be re-acquired. A more ecient approach to this problem is to acquire motion corrupted data and attempt to correct this data afterwards.

Novel approaches for respiratory motion correction are developed in this thesis. The proposed framework estimates complex, non-rigid motion from the data itself. The motion information is then incorporated into the reconstruction to remove motion-related artefacts. This non-rigid motion correction framework is adapted to three different applications: 3D accelerated abdominal imaging, 3D coronary lumen and vessel wall imaging, and 3D whole-heart water/fat imaging. In the rst application, the framework is combined with Parallel Imaging and Compressed Sensing to enable high acceleration factors. The proposed method reduced scan times by 2.6x when compared with the gated acquisition while maintaining similar image quality. In the second application, the framework is combined with interleaved image navigators to add high temporal resolution motion correction. This method also presented similar coronary lumen quality to the gated, despite a 1.6x reduction in scan time. Additionally, it presented signi cantly superior vessel wall quality when compared to translation correction. In the third application, the framework is combined with Parallel Imaging, Compressed Sensing and interleaved image navigators. Initial results indicate the proposed approach produces signi - cantly superior water and fat images than translation correction.
Date of Award2016
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorClaudia Prieto Vasquez (Supervisor) & Tobias Schaeffter (Supervisor)

Cite this