Multi-site and Multi-vendor T2 Mapping from Proton Density and T2-Weighted Images

  • Author / Creator
    Chhetri, Gitanjali
  • Transverse relaxation (T2)-weighted images are routinely acquired in clinics. However, T2 quantification is rarely performed as most of the specialized pulse sequences used to quantify T2 are time consuming and thus generally not used in a clinical setup. Nevertheless, there is strong interest in quantitative measures in research settings. T2 quantification provides an unbiased measurement of tissue characteristics; it has been shown to be sensitive to a wide range of brain pathologies like Dementia, Multiple Sclerosis, Stroke, and Epilepsy, to name a few. T2 measurement is commonly performed by curve-fitting an exponential decay to a long spin-echo train typically acquired from multiple-echo spin-echo sequences. However, errors arise from imperfect refocusing due to imperfect slice profiles and radio frequency interference effects at 3 T and higher fields. Imperfect refocusing leads to contamination of the T2 decay curve by contributions from stimulated and indirect echoes. Bloch equation-based modelling of the multi-echo spin-echo sequences have shown to provide accurate T2 measurements when a flip angle map is provided to the fitting method. Large group MRI studies allow comparison of data across multiple sites and vendors, provided that the pulse sequences are standardized between the scanner vendors and also between different scanner versions. This thesis is focused on testing the viability of using a T2 quantification method that can be applied across vendors and sites on existing clinical images. The method must make use of already acquired clinical images to ultimately enable clinical use without excessive sequence additions. Here we consider the proton-density and T2-weighted dual echo images for such T2 measurement. We examine retrospective T2 mapping across multiple sites and two vendors using only proton density and T2-weighted fast spin-echo images and an included calibration scan using the Alzheimer’s Disease Neuroimaging Initiative (ADNI-1) database for healthy subjects at 3 T. We show that a simple two-point exponential fit of these two images leads to striking biases in T2 measurements between vendors owing to differences in pulse sequence parameters. By using Bloch-based modelling of the pulse sequence and estimated B1+ maps, these biases were removed. The outcome of this work is that T2 quantification is possible from standard clinical images across sites and vendors if one models the actual pulse sequence and flip angles applied.

  • Subjects / Keywords
  • Graduation date
    Spring 2021
  • Type of Item
  • Degree
    Master of Science
  • DOI
  • License
    This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for non-commercial purposes. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.