Optimization of a 48 echo magnetic resonance imaging sequence using variable TR data acquisition

Abstract

Magnetic resonance imaging (MRI) is a very valuable tool for studying the brain. Currently, MRI is the only non-invasive method for investigating myelin. A unique MRI pulse sequence which is used to investigate myelin is a 48 echo CPMG experiment with TR = 3800ms and T E 10 and 50ms. Unfortunately, this experiment takes over 33 minutes to complete, making clinically less feasible to use. By collecting higher order regions of k-space at shorter TR times, the experiment can be shortened, but at a cost of increasing image blurrines and at a potential loss of data. The purpose of this thesis was to investigate collecting different regions of k-space at different TR times in order to try and optimize a new 48 echo variable TR pulse sequence. Simulations were first performed using five spin-echo images with different TR. By creating simulated variable TR images, we were able to qualitatively investigate the resulting blurriness of the images. Visual assesment of the created images and the difference images allowed us to determine what degree of resolution deterioration would still allow us to differentiate between important structures. It was decided that the simulation for 60 out of 128 lines collected at a shorter TR had the optimal decrease in scan time, without too great a compromise in image quality. The variable TR C P M G experiment was then run on 9 phantoms with different Tr and T2 relaxation times. By studying samples with known T, and T 2 relaxation times, we were able to investigate the reliability of the variable TR pulse sequence. Comparing decay curves showed no difference between 0 and 100 lines of k-space collected at a shorter TR it was only when all 128 lines of kspace were collected at the shorter TR that a decrease in amplitude of the decay curve occurred. Experiments showed that proton density, GMT2 and chi squared of the T 2 decay curve fit for the phantoms were unaffected up to and including 100 lines of k-space collected at TR of 2120ms. Finally, in-vivo studies were performed on five volunteers. Comparing the difference in decay curves, proton density and geometric mean T 2 showed only very minor differences between data collected using the constant TR sequence and data collected using the variable TR program in which 60 out of 128 k-space lines were collected at a shorter TR of 2120ms. Experiments showed small differences in myelin water fraction, which could be explained by ROI's being drawn slightly different on the constant and variable images. The chi squared was less for the variable TR, which could be caused by smoothing introduced when collecting different k-space lines at different TR's.