Dear all, 

I'm using DTIPrep and after transposing the bvec (3 columns) and bval (one column) files, it gives the correct appearance in the DTIPrep. However, after running the QC, it makes all the gradients excluded except of the baseline images.

In the text file it comes out like this:
Included gradients:
Baselines_indices: 0 QCIndex: 0
Baselines_indices: 1 QCIndex: 0
Baselines_indices: 2 QCIndex: 0
Baselines_indices: 3 QCIndex: 0
Baselines_indices: 4 QCIndex: 0
Baselines_indices: 5 QCIndex: 0

Does somebody know how to solve this problem?
Thanks in advance.

I think we need a bit more information. 
1. Which QC step rejects all the non-baseline gradients? (look at the report text file) What are the protocol settings for that step?
2. Since you started with a nifti file, you had to first convert it to NRRD? If so, does the header information look correct (the NRRD header is ASCII, so you can look at it with any editor)? Do the gradient directions look correct in the NRRD header?

Martin

Dear Martin, 

Thank you for your quick response!
1. It is the gradient-wise checking (see attached QC report). 
Parameters:
TranslationThreshold: 1.31773
RotationThreshold: 0.5
2. The gradient directions look ok when loading the nrrd-file into the DTIPrep GUI. We've used DWIConvert from the command line and not created separate header files. When I convert in Slicer and save a separate header file, the gradient directions have a different precision. Might this have to do with which space the data is in? (LPS vs. RAS)? The contents of the .nhdr are found below (since I can't attach two files). Anyway, it looks ok in DTIPrep.

It should perhaps be mentioned that we're processing older data (~10 years).

Best,
Karen


NRRD0005
# Complete NRRD file format specification at:
# http://teem.sourceforge.net/nrrd/format....
type: short
dimension: 4
space: right-anterior-superior
sizes: 38 256 256 64
space directions: none (-0.97659999132156361,0,0) (0,-0.97659999132156361,0) (0,0,2)
kinds: list domain domain domain
endian: little
encoding: gzip
space origin: (124.70899963378905,134.45399540662763,-79.152496337890625)
measurement frame: (-1,0,0) (0,-1,0) (0,0,1)
data file: test.raw.gz
DWMRI_b-value:=1000.000000
DWMRI_gradient_0000:=0.000000 0.000000 0.000000
DWMRI_gradient_0001:=0.000000 0.000000 0.000000
DWMRI_gradient_0002:=0.000000 0.000000 0.000000
DWMRI_gradient_0003:=0.000000 0.000000 0.000000
DWMRI_gradient_0004:=0.000000 0.000000 0.000000
DWMRI_gradient_0005:=0.000000 0.000000 0.000000
DWMRI_gradient_0006:=-1.000000 0.000000 0.000000
DWMRI_gradient_0007:=-0.494198 -0.869349 0.000000
DWMRI_gradient_0008:=0.120009 -0.641050 0.758059
DWMRI_gradient_0009:=-0.919038 0.331014 -0.214009
DWMRI_gradient_0010:=0.111970 0.630833 0.767797
DWMRI_gradient_0011:=0.914305 0.236079 0.329110
DWMRI_gradient_0012:=-0.680844 -0.302930 0.666847
DWMRI_gradient_0013:=-0.066978 -0.997670 -0.012996
DWMRI_gradient_0014:=-0.917623 -0.152937 0.366849
DWMRI_gradient_0015:=-0.443231 0.890464 0.103054
DWMRI_gradient_0016:=-0.215975 0.926892 -0.306964
DWMRI_gradient_0017:=-0.480032 0.671045 -0.565038
DWMRI_gradient_0018:=0.549943 0.477950 0.684928
DWMRI_gradient_0019:=0.700688 -0.692691 0.170924
DWMRI_gradient_0020:=0.429108 0.007002 -0.903226
DWMRI_gradient_0021:=0.698790 0.639808 0.319904
DWMRI_gradient_0022:=-0.725825 -0.090978 -0.681836
DWMRI_gradient_0023:=-0.389830 0.248892 -0.886614
DWMRI_gradient_0024:=0.172056 0.881289 -0.440144
DWMRI_gradient_0025:=0.609674 -0.640657 -0.466750
DWMRI_gradient_0026:=-0.598760 -0.688724 0.408836
DWMRI_gradient_0027:=0.727829 -0.221948 -0.648848
DWMRI_gradient_0028:=-0.882959 0.397981 0.248988
DWMRI_gradient_0029:=-0.320210 -0.548360 0.772507
DWMRI_gradient_0030:=0.044018 0.268107 -0.962383
DWMRI_gradient_0031:=-0.188896 0.876518 0.442757
DWMRI_gradient_0032:=-0.732165 0.333075 -0.594134
DWMRI_gradient_0033:=-0.308976 0.549957 0.775939
DWMRI_gradient_0034:=-0.285959 -0.857876 -0.426938
DWMRI_gradient_0035:=-0.859154 -0.504090 0.088016
DWMRI_gradient_0036:=0.106972 -0.214945 -0.970750
DWMRI_gradient_0037:=0.297951 0.177970 0.937844
modality:=DWMRI
space:=left-posterior-superior

Thanks Karen

The gradient-wise checking uses an alternative registration to detect alternative motion and is a bit more susceptible to low SNR data (and overestimates particularly residual translation motion in such a case). I would disable the gradient-wise/residual motion check phase (set it to "No" in the protocol), as it is not stable with your data.

re header: If only the precision is different, than that should be fine, you have plenty of precision in the gradient directions in that header. 

one more note: your DTI data is not isotropic (0.98x0.98x2), which is not recommended. After DTIPrep, I would first resample the DWI data to 0.98x0.98x0.98 and only then fit the tensor etc

Best
Martin
Originally posted by karen:

Hi Martin, 

Thanks a lot for your suggestions!

Best, 
Karen