Work Description

Date written: June 8, 2025
README Author: Amaya Murguia
README Author Email: [email protected]

Data accompanying paper: "Impact of Tissue Sample Preparation Methods on Myelin-Sensitive Quantitative MR Imaging"
Citation: Murguia, A., Swanson, S. D., Scheven, U., Jacobson, A., Nielsen, J. F., Fessler, J. A., & Seraji-Bozorgzad, N. (2025).
Impact of Tissue Sample Preparation Methods on Myelin-Sensitive Quantitative MR Imaging. bioRxiv, 2025-05.
https://www.biorxiv.org/content/10.1101/2025.05.05.652075v1.abstract

Research project description: Validation of quantitative MRI (qMRI) parameters with histology is often done with ex vivo fixed tissue samples.
Freezing is another common form of tissue preservation, but the effects of freezing and thawing tissue on myelin-sensitive qMRI parameters
and their correlation with histology require further analysis. To study the effects of freezing vs. fixing tissue on myelin-sensitive parameters,
we conducted myelin water imaging, off-resonance RF saturation magnetization transfer (MT), and selective inversion recovery MT MRI experiments
on 14 fresh, thawed, and fixed sheep brain tissue samples to calculate various surrogate measures of myelin content. These measures were
compared with luxol fast blue (LFB) histological stain results. 14 sheep brain tissue samples were scanned fresh.
Half were frozen and subsequently thawed and scanned again. The second half were fixed and scanned again.
Then all samples were fixed and sent for histological processing.

Methods: MRI scans of 14 sheep brain tissue samples were acquired on a 7.0 Tesla NMR/MRI scanner (Varian/Agilent).
Various scans were acquired: gradient echo, fast spin echo, multi-echo spin echo, B0 mapping, B1 mapping, inversion recovery,
and magnetization transfer scans. Histology (Luxol fast blue stain / Cresyl violet counterstain) images were also acquired.

The top level folder sheep_data_paper has subfolders:
fresh_thawed: holds the data for the samples scanned fresh and thawed (S1-S7)
fresh_fixed: holds the data for the samples scanned fresh and fresh_fixed (S8-S14)

Each structure variable (data.mat) holds data for a specific sample and a specific tissue preparation (fresh, thawed, or fixed).
The data variable has the following fields:

gre: gradient echo scan data (128 x 128 x 5)
fse: fast spin echo scan data (256 x 256 x 5)
mese: multi-echo spin echo scan data for MWF mapping (128 x 128 x 5 x 104)
b0data: gradient echo scan data for B0 mapping (128 x 128 x 5 x 4)
b1data: EPI scan data for B1 mapping (32 x 32 x 5 x 9)
t1data: EPI inversion recovery data for T1 mapping (64 x 64 x 5 x 21)
mtdata: gradient echo data for quantitative MT analysis (128 x 128 x 4 x nf)
hist: Luxol fast blue stain / Cresyl violet counterstain (LFB) histology image

This folder contains the sheep brain tissue MR and histology images collected for this study.
All data are stored as mat files.

top level folder name: sheep_data_paper
subfolders:
fresh_thawed: holds the data for the samples scanned fresh and thawed (S1-S7)
fresh_fixed: holds the data for the samples scanned fresh and fresh_fixed (S8-S14)

Each structure variable (data.mat) holds data for a specific sample and a specific tissue preparation (fresh, thawed, or fixed).
The data variable has the following fields:

gre: gradient echo scan data (128 x 128 x 5)
fse: fast spin echo scan data (256 x 256 x 5)
mese: multi-echo spin echo scan data for MWF mapping (128 x 128 x 5 x 104)
b0data: gradient echo scan data for B0 mapping (128 x 128 x 5 x 4)
b1data: EPI scan data for B1 mapping (32 x 32 x 5 x 9)
t1data: EPI inversion recovery data for T1 mapping (64 x 64 x 5 x 21)
mtdata: gradient echo data for quantitative MT analysis (128 x 128 x 4 x nf)
hist: Luxol fast blue stain / Cresyl violet counterstain (LFB) histology image

Each of the variables (with MR data) is also a structure variable with the following fields:
image (complex image data)
kspace (complex kspace data, only have for the non-EPI data - gre, fse, mese, b0data, mtdata)
pars (parameters pulled off the scanner)
acquisitiontype (2D or 3D)

hist is also a structure with the following fields
LFB_RGB (uint8 of size 257 x 173 x 3, downsampled LFB image)
LFB_B_minus_R (uint8 of size 257 x 173 with the blue minus the red channel of the downsampled RGB LFB image)

Scan parameters (please see the paper for more details):

Slice plan: 5 slices, interleaved, 2mm thickness, 0mm gap

Non-EPI scan details
gre: TR = 100ms, TE = 4ms, matrix size = 128 x 128, 5 slices
fse: TR = 4000ms, TE (effective) = 40ms, echo spacing = 10ms, segments/echo train length = 32/8, k-zero = 4, averages = 1, matrix size = 256 x 256
mese: TR = 4000ms, TE = 5ms, number of echoes = 104, averages = 2, matrix size = 128 x 128
b0data: TR = 100ms, TE array of 4, 6, 8, 10ms, flip angle = 20 deg, averages = 1, matrix size = 128 x 128
mtdata: 1 slice (not 5 for this scan), TR = 120ms, TE = 3ms, flip angle = 20 deg, averages = 1, matrix size = 128 x 128, 4 power levels (satpwr),
nf off-resonance frequencies (satfrq1) (nf = 25 usually, for a few samples nf = 21)

EPI scan details
b1data: TR = 8000ms, TE = 9.55ms (minimum), flip angle = array of 9 values from 20-180 deg spaced 20 deg apart, averages = 1 (1 shot, 1 repetition), matrix size = 32 x 32
t1data: TR = 8000ms, TR = 36.64ms (minimum), averages = 1 (1 shot, 1 repetition), matrix size = 64 x 64, spin echo, TI array of 21 inversion times (log spaced from 10ms-5ms)

Useful parameter names from pars file (these variable names are the Varian scanner parameter names):
ns: number of slices
te: TE in seconds
tr: TR in seconds
ne: number of echoes
flip1: flip angle in degrees
ti: TI (inversion time) in seconds
resto: resonance transmitter offset frequency in Hz
thk: slice thickness in millimeters
pw: pulse width in microseconds
satfrq1: off-resonance frequency in Hz (for MT data preparation pulse)
satpwr: power level in dB (for MT data preparation pulse)

Not every sample has data for every MR measure; please see Table S1 in the paper.