• Login
• Admin

Bortolotti, Laura (2022) Monitoring head motion in a 7T MRI scanner using an NMR field camera. PhD thesis, University of Nottingham.

Preview

PDF (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (116MB) | Preview

Abstract

Magnetic Resonance Imaging (MRI) is a technique for imaging the soft tissues of the human body. Research in medical MRI is moving towards the use of ultra-high field (UHF) MRI scanners since the spatial resolution of MRI increases with the strength of the magnetic field. However, involuntary movements of the subject create artefacts that can corrupt MR images. Since high resolution images are more vulnerable to motion, this effect becomes more relevant at ultra-high field. Motion artefacts can be ameliorated if the movements of the head during the scans are tracked; this is the basis of motion correction (MoCo) techniques.

This thesis explores a novel way to approach the motion tracking step of MoCo techniques in a marker-less way. The core idea is to measure the extra-cranial magnetic field changes produced by changes in head pose, and then to use these measurements to infer information about head motion. The former was measured using a magnetic field camera, a fully MRI compatible tool. In a final implementation of the approach, the field probes were held on the surface of the receiver RF coil of a 7 T scanner using a hand-made probe holder. Measurements were then acquired in the quiet periods of a multi-slice echo planar imaging sequence.

Simultaneous measurements of extra cranial magnetic field changes and head motion parameters were acquired while six collaborative subjects were instructed to perform several different types of head movement inside the scanner. The Moire’ Phase Tracking System has been considered as the gold standard for evaluating the six head motion parameters inside the scanner bore.

A spatial filter, based on the use of solid harmonic functions, has been developed in order to reduce the influence of the field changes due to physiological fluctuations. Feature selection on magnetic field data has been computed using Principal Component Analysis. The subgroup of signals were identified by applying Hierarchical Cluster Analysis to signals projected into the principal component space. Furthermore, field changes due to physiological fluctuations have been exploited to generate a simple signal that could be used for respiratory monitoring.

Customised extra-cranial magnetic field simulations were implemented to test the spatial filtering process and the feasibility of using extra-cranial magnetic field changes to track head movements. Head motion parameters were predicted from simulated extra cranial magnetic field changes by using linear and a non-linear regression methods, both previously trained by supervised learning. The linear method chosen was the Partial Least Squares method. The non-linear method chosen was a single hidden layer, recurrent and dynamic neural network based on Non-linear AutoRegressive eXogenous model (NARX). Results obtained using simulated data were confirmed on experimental data over different subjects, head motion ranges, probe displacements and sessions, using raw and spatially filtered data. A preliminary study on generalising the regression method over twenty subjects has also been conducted on simulated data.

Furthermore, extra-cranial magnetic field changes were used to discriminate between predictable and non-compensable head movements. The former are head movements well predicted by the head motion tracking (either the one developed in this dissertation or not). The latter are head movements that cannot be well compensated by applying MoCo techniques to MRI. Thus, they might lead to the need to repeat the whole MRI acquisition. The solution suggested in this dissertation is to use information given by the magnetic field probes to flag the k-space lines that need to be re-acquired due to motion effects. This could save scanning time and reduce patient discomfort.

A pilot study of an active-magnetic marker based system has been carried out using customised simulations. This system would rely on the use of the NMR field probes to detect a local magnetic field generated by small coils (0.5 cm of diameter) fixed on a pair of plastic glasses. Thanks to the use of an optimisation algorithm, the position of the coil-based system is resolved. In its future development, this system might substitute the use of the optical camera as a stand-alone system or for the purpose of the training of the motion tracking system developed in this dissertation.

Results reported in this thesis represent a step towards the full development of a marker-less technique for head motion tracking that does not require modification of the MR image sequence. In its future development, this technique can be used to improve the outcome of standard MRI procedures.

Item Type:	Thesis (University of Nottingham only) (PhD)
Supervisors:	Bowtell, Richard Gowland, Penny
Keywords:	Magnetic Resonance Imaging, Brain imaging, Ultra high magnetic field, Motion correction, Data analysis, Machine learning
Subjects:	Q Science > QC Physics > QC501 Electricity and magnetism
Faculties/Schools:	UK Campuses > Faculty of Science > School of Physics and Astronomy
Item ID:	68725
Depositing User:	Bortolotti, Laura
Date Deposited:	02 Aug 2022 04:40
Last Modified:	02 Aug 2022 04:40
URI:	https://eprints.nottingham.ac.uk/id/eprint/68725

Actions (Archive Staff Only)

Edit View