Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Medical Biophysics


Dr. Stewart Gaede


The hybridization of magnetic resonance imaging (MRI) and positron emission tomography (PET) provides the benefit of soft-tissue contrast and specific molecular information in a simultaneous acquisition. The applications of PET-MRI in radiotherapy are only starting to be realised. However, quantitative accuracy of PET relies on accurate attenuation correction (AC) of, not only the patient anatomy but also MRI hardware and current methods, which are prone to artefacts caused by dense materials. Quantitative accuracy of PET also relies on full characterization of patient motion during the scan. The simultaneity of PET-MRI makes it especially suited for motion correction. However, quality assurance (QA) procedures for such corrections are lacking. Therefore, a dynamic phantom that is PET and MR compatible is required. Additionally, respiratory motion characterization is needed for conformal radiotherapy of lung. 4D-CT can provide 3D motion characterization but suffers from poor soft-tissue contrast. In this thesis, I examine these problems, and present solutions in the form of improved MR-hardware AC techniques, a PET/MRI/CT-compatible tumour respiratory motion phantom for QA measurements, and a retrospective 4D-PET-MRI technique to characterise respiratory motion.

Chapter 2 presents two techniques to improve upon current AC methods that use a standard helical CT scan for MRI hardware in PET-MRI. One technique uses a dual-energy computed tomography (DECT) scan to construct virtual monoenergetic image volumes and the other uses a tomotherapy linear accelerator to create CT images at megavoltage energies (1.0 MV) of the RF coil. The DECT-based technique reduced artefacts in the images translating to improved μ-maps. The MVCT-based technique provided further improvements in artefact reduction, resulting in artefact free μ-maps. This led to more AC of the breast coil.

In chapter 3, I present a PET-MR-CT motion phantom for QA of motion-correction protocols. This phantom is used to evaluate a clinically available real-time dynamic MR images and a respiratory-triggered PET-MRI protocol. The results show the protocol to perform well under motion conditions. Additionally, the phantom provided a good model for performing QA of respiratory-triggered PET-MRI.

Chapter 4 presents a 4D-PET/MRI technique, using MR sequences and PET acquisition methods currently available on hybrid PET/MRI systems. This technique is validated using the motion phantom presented in chapter 3 with three motion profiles. I conclude that our 4D-PET-MRI technique provides information to characterise tumour respiratory motion while using a clinically available pulse sequence and PET acquisition method.