Ex-vivo MRI data were acquired on a 7T 30cm (Bruker Biospin) MRI platform with a Bruker 35mm linear birdcage volume coil. Before MRI scanning, formalin-fixed marmoset brains were soaked with 0.15% gadopentetate dimeglumine (Magnevist, Bayer, Leverkusen, Germany) for 3 weeks to reduce the T1 relaxation time. The brains were fitted into 3D-printed brain holders to maintain them in an orientation convenient for later data processing. Multi-modal MRI data were collected for one brain sample, including multi-shell diffusion MRI, MTR, and T2w image: The multi-shell diffusion MRI was collected with a 3D diffusion-weighted multi-shot spin-echo EPI sequence: TR = 450ms, TE = 34ms, FOV = 38.4×28.8×28.8 mm, matrix size = 256x192x192, resolution = 0.15 mm isotropic, a total of 396 DWI images for 3 shells (b = 2400, 4800, and 7200), 6 b0 and 126 directions per shell, number of average = 1 for the shells b= 2400 and 4800, number of averages = 2 for the shell b= 7200, and total acquisition time was about 150 hours (more than 6 days). Multi-shell gradient sampling schemes were generated by the IMOC method of the DMRITool (Cheng et al., 2017). The MTR image was collected with a 3D FLASH sequence: TR = 21.5 ms, TE = 3.6 ms, flip angle = 20o, FOV = 38.4×28.8×28.8 mm, matrix size = 256x192x192, resolution = 0.15 mm isotropic, number of averages = 4 and number of repetitions = 5, with (Ms) and without (M0) an offset magnetization transfer (±2000 HZ off resonance, Gaussian shaped), and total acquisition time was about 17.6 h. The T2w image was collected with a 3D MSME sequence: TR = 2000 ms, TE = 10.4 ms, FOV = 38.4×25.5×25.5 mm, matrix size = 256x170x170, resolution = 0.15 mm isotropic, number of averages = 1, and total acquisition time was about 13h.
Download Raw DWI data*We masked out most black background and converted the data to 16bit integer format (.nii.gz) to reduce the size of the data. *The b7200 had two "number of averages" while other b values (including b0) only had one "number of averages". When the Bruker scanner (PV5) acquired the data, the "average" was more like the "sum" of two acquisition of signals, as the Bruker scanner (PV5) didn't divide the value by 2. Thus, we divided the b7200 of raw_noscale.nii.gz by a factor of 2 to make the value-scaling constant across different shells. We released the scaled raw data (raw.nii.gz) here. If you want the raw_noscale.nii.gz (for denoising purpose for example), you can extract the b7200 of raw.nii.gz and multiple it by 2. *Structural images (MTR and T2w) were provided with the downloading package of the Preprocessed DWI data (see below).
Required software: Mrtrix3, FSL
1. Denoising DWI (by Mrtrix3)
Note: The b7200 had two "number of averages" while other b values (including b0) only had one "number of averages". When the Bruker scanner (PV5) acquired the data, the "average" was more like the "sum" of two acquisition of signals, which casued different scaling effect on the signal and noise values.Thus, we performed the following steps for the DWI denoising. More details discussion can be found in this link.
$ mrconvert raw_noscale.nii.gz raw_noscale.mif -fslgrad raw.bvecs raw.bvals
$ dwiextract raw_noscale.mif -shell 7200 b7200.mif
$ dwiextract raw_noscale.mif -shell 0,2400,4800 raw_other_shells.mif
$ mrcalc b7200.mif 2 -sqrt -div b7200_sqrt2.mif
$ mrcat raw_other_shells.mif b7200_sqrt2.mif -axis 3 dwi_constant_noise.mif
$ dwidenoise dwi_constant_noise.mif dwi_denoised.mif -noise dwi_noise.nii.gz -mask raw_mask.nii.gz -extent 9,9,9 -force
$ dwiextract dwi_denoised.mif -shell 7200 b7200_denoised.mif
$ mrcalc b7200_denoised.mif 2 -sqrt -div b7200_denoised_sqrt2.mif
$ dwiextract dwi_denoised.mif -shell 0,2400,4800 other_shells.mif
$ mrcat other_shells.mif b7200_denoised_sqrt2.mif -axis 3 dwi_final.mif
$ mrconvert dwi_final.mif dwi_final.nii.gz -export_grad_fsl dwi_final.bvecs dwi_final.bvals
2. Eddy current correction and DTI-fitting (by FSL)
$ eddy_correct dwi_final.nii.gz data.nii.gz 0 spline
$ fdt_rotate_bvecs dwi_final.bvecs bvecs data.ecclog
$ dtifit -k data.nii.gz -o DTIFIT -m mask.nii.gz -r bvecs -b bvals
*We masked out most black background and converted the data to 16bit integer format (.nii.gz) to reduce the size of the data. *Structural images (MTR and T2w) is included here.