Myelin water fraction in post-mortem brains: sequence comparison and behavior through fixation time
Poster No:
2383
Submission Type:
Abstract Submission
Authors:
Liana Sanches1, Roqaie Moqadam1, Mallar Chakravarty2, Mahsa Dadar3, Yashar Zeighami4
Institutions:
1Cerebral Imaging Center, Douglas Mental Health University Institute, Montreal, Quebec, 2Brain Imaging Centre, Douglas Research Centre, Montreal, Quebec, 3Douglas Research Centre, McGill University, Montreal, QC, 4Douglas Research Centre, McGill University, Montreal, Quebec
First Author:
Liana Sanches, PhD
Cerebral Imaging Center, Douglas Mental Health University Institute
Montreal, Quebec
Cerebral Imaging Center, Douglas Mental Health University Institute
Montreal, Quebec
Co-Author(s):
Introduction:
The use of Magnetic resonance imaging (MRI) in neuroscience goes beyond the investigations of brain structure. Quantitative magnetic resonance imaging (qMRI) sequences provide an indirect proxy for tissue microstructure. Myelin water fraction (MWF) is a multi-compartmental modeling approach that uses the T2 relaxation time to calculate the amount of water trapped in the myelin sheets, which provides an indirect estimate of white matter integrity (Laule et al, 2004). For a method to be practical and feasible, all of this has to be accomplished in tolerated acquisition times and be applicable in clinical scanners. Studies in post-mortem MR of brain samples can help achieve these goals and validate the relationship between obtained MWF estimates and real microstructural structures. Finally, they can be compared with in vivo antemortem data and histology analysis. Therefore, post-mortem investigations can provide the ideal setting to develop and validate new biomarkers for neurodegenerative disorders (Shatil et al, 2018). Here we compare the standard 2D T2 (turbo) multi spin-echo (Neumann et al, 2014) and the 3D gradient and spin echo sequence (GRASE) (Piredda et al, 2020) acquisitions, in post-mortem brain samples fixed for different times. We calculate the correlation of MWF between the sequences and their behavior through the time since fixation.
Methods:
Brain samples: Data included 12 post-mortem brain hemispheres from the Douglas Bell Canada Brain Bank (5 female, mean age = 80.7± 0.8). The donated brains were cut at the midsection, and one hemisphere (left or right, randomly assigned) was fixed in a 10% buffered formalin solution. The brains were scanned at different fixation times to allow us to assess the impact of fixation on the qMRI results (fixation time ranging between 0-20 years). All images were acquired at the Douglas CIC Research Centre, in a 3T scanner with 64 channel head-neck coil. The GRASE is a "work-in-progress" sequence obtained through a research agreement with the vendor. The following parameters were used: voxel size: 1.8 mm3, repetition time (TR): 1000 ms, 1 average, acceleration mode factor 3 in phase-encoding direction, and 2 in 3D. 32 echo times (TE) from 9 to 288 ms, echo-spacing 0.94 ms. The turbo spin-echo sequence has 27 slices and voxels of 0.3 x 0.3mm with 2.5 mm thickness, TR=4300 ms, 18 echo train per slice, turbo factor of 8, acceleration mode factor 3 in phase-encoding direction. This sequence was performed 7 times, changing the TE from 8 to 58 ms (echo spacing 8.25 ms). BISON (Dadar and Collins, 2021), a tissue segmentation tool was used to acquire white matter masks (WM) based on T1w images. The tissue masks were resampled to the spin-echo and GRASE images, providing regions of interest in which MWF and T2 values of the two sequences were compared. We assessed the correlation between average WM MWF estimations based on the two sequences, as well as the average WM MWF and fixation time, including post-mortem delay, sex, age, and hemisphere as covariates.
Results:
Figure 1 shows an example of the derived MWF maps for one specimen, demonstrating the overall spatial agreements between the maps derived from the two sequences. Average white matter MWF estimations from the two sequences were highly correlated with each other (r = 0.90, p=0.0001), and positively correlated with fixation time (r = 0.78, p = 0.005 for standard T2 multi spin-echo and r = 0.63, p = 0.03 for GRASE). However, the average MWF estimates were consistently lower based on the multi-spin echo compared to the GRASE (t stat = -4.65, p<0.0001).
·Figure 1. Comparison of T2 relaxation maps derived from GRASE and standard T2 multi spin echo sequences. a. post-mortem T1w image. b-d. Corresponding BISON tissue segmentation, GRASE MWF, and standard
Conclusions:
Our results showed that the GRASE sequence can provide a reliable proxy for the standard T2 multi spin-echo sequence, with the advantage of a better isotropic spatial resolution, brain coverage, and lower acquisition times (9:28 minutes compared to 15).
Modeling and Analysis Methods:
Image Registration and Computational Anatomy 2
Novel Imaging Acquisition Methods:
Imaging Methods Other 1
Keywords:
Acquisition
MRI
MRI PHYSICS
Myelin
Segmentation
White Matter
Other - T2 Relaxation; postmortem
1|2Indicates the priority used for review
Provide references using author date format
Laule C, (2004), 'Water content and myelin water fraction in multiple sclerosis. A T2 relaxation study.' Journal of neurology, Mar;251(3):284-93.
Neumann D, (2014), 'Simple recipe for accurate T(2) quantification with multi spin-echo acquisitions.' Magnetic Resonance Materials in Physics, Biology and Medicine. Dec;27(6):567-77.
Piredda, GF. (2020), 'Fast and high-resolution myelin water imaging: Accelerating multi-echo GRASE with CAIPIRINHA'. Magnetic Resonance in Medicine,; 85: 209–222.
Shatil AS. (2018), 'Quantitative Ex Vivo MRI Changes due to Progressive Formalin Fixation in Whole Human Brain Specimens: Longitudinal Characterization of Diffusion, Relaxometry, and Myelin Water Fraction Measurements at 3T. Frontiers in Medicine. 5:31.