Volumes lost after dwi2fod

1

Dear MRtrixer,
I have mouse diffusion MRI data 5 b-values (0, 630, 1230, 1831, 2433) with 8, 18, 18, 36, 36 volumes. I run the dwi2fod command and see the output fod file contains only 44 volumes. I guess they are from first three shells (0, 630 and 1230). I am wondering where are the other 72 volumes from shell 1831 and 2433.

The following is white matter fod information
mrinfo.exe wm-fod.mif

Image name: “wm-fod.mif”

Dimensions: 100 x 86 x 54 x 45
Voxel size: 0.15 x 0.15 x 0.3 x 1
Data strides: [ 2 3 4 1 ]
Format: MRtrix
Data type: 32 bit float (little endian)
Intensity scaling: offset = 0, multiplier = 1
Transform: 1 0 0 0
0 1 0 0
0 0 1 0
command_history: C:\msys64\mingw64\bin\dwi2fod.exe msmt_csd dwi-processed.nii -fslgrad bvec.bvec bval.bval wm.txt wm-fod.mif gm.txt gm-fod.mif csf.txt csf.mif -nthreads 8 (version=3.0.3)
comments: 6.0.5:9e026117
mrtrix_version: 3.0.3
prior_dw_scheme: 0,0,0,0
[116 entries] 0,0,0,0
…
0.3268800453,0.930013129,0.1680030233,1819.109495
0.3268800453,0.930013129,0.1680030233,2415.45933

The following is wm-response information
$ cat wm.txt
# Shells: 0,630,1229,1831,2433
# command_history: C:/msys64/mingw64/bin/amp2response.exe dwi.mif voxels_sfwm.mif safe_vecs.mif response_sfwm.txt -shells ‘0,630,1230,1831,2433’ -nthreads 8 (version=3.0.3)
9812.2302254316 0 0 0 0 0
6484.73088705417 -1330.3019937495 178.870372147496 7.30039661381014 19.3939997016384 -40.6092939720243
4936.20709899322 -1483.75337756058 358.077762801241 -52.5880645369274 11.4483335336519 5.96712556968778
6290.63447824885 -2006.41723343585 793.925220816595 -212.394065545424 14.0879694792254 8.58673222393911
5461.94704942416 -1778.95773842118 833.949139610945 -267.671642906618 75.0197704445543 -17.9206319570738

The following is original diffusion image information
mrinfo.exe dwi.nii.gz

Image name: “dwi.nii.gz”

Dimensions: 100 x 86 x 54 x 116
Voxel size: 0.15 x 0.15 x 0.3 x 1
Data strides: [ 1 2 3 4 ]
Format: NIfTI-1.1 (GZip compressed)
Data type: signed 16 bit integer (little endian)
Intensity scaling: offset = 0, multiplier = 1
Transform: 1 0 0 0
0 1 0 0
0 0 1 0
comments: untitled
mrtrix_version: 3.0.3-98-gb3409159

Any solution? Thanks in advance.

1 Like
2

Hello!

Don’t worry, nothing has gone wrong here :slight_smile: This is the expected output for dwi2fod.
dwi2fod performs constrained spherical deconvolution (CSD) to estimate fibre orientation distributions (FODs) from diffusion data. The output (wm-fod.mif) is no longer shelled data where each volume corresponds to a different bvalue/bvec combination – it’s an entirely new type of image which encodes a distribution on a sphere.

In MRtrix3, such distributions are represented in spherical harmonic (SH) basis. In your wm-fod.mif image, each volume corresponds to an SH coefficient. The number of coefficients (and thus volumes) in an SH image is determined by the maximum order l_{max} used to represent the distribtution (simply put, the higher the value of l_{max}, the more angular detail). If l_{max}=8 (which is the highest value and the default in your case) then that corresponds to 45 image volumes.

So the number of volumes in the output of dwi2fod is not determined by the number of volumes in the input diffusion image, just like how you can model the diffusion tensor from 6 or 30 directions, and the output will always be a symmetrical 3x3 tensor (or 6 unique parameters stored in 6 volumes in MRtrix3). In both cases, the extra input data (beyond the minimum required) improves your model fitting / conditioning, but doesn’t increase the number of parameters in the output.

If you need more details, all this is explained much better by the experts
Spherical Harmonics — MRtrix 3.0 documentation.
Maximum spherical harmonic degree lmax — MRtrix 3.0 documentation.
Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution - PubMed.
Multi-tissue constrained spherical deconvolution for improved analysis of multi-shell diffusion MRI data - PubMed.

Hope that helps!

Fiona

4 Likes
3

I do understand now.
Thanks for your kind reply.