Error when converting t1w to BIDS: t1w file with too many dimensions

,
1

Hi everyone,

I’m trying to convert my MRI (mostly t1w) database into BIDS, from the nifti files (i don’t have any DICOM).
For half of the subjects (130), i’m facing an error when i try to validate my dataset using BIDS validator:

Error 3: [Code 95] T1W_FILE_WITH_TOO_MANY_DIMENSIONS.
sub-0XXX_ses-1_T1w.nii.gz
11054.205 KB | application/gzip
Location:
BIDS/sub-0XXX/ses-1/anat/sub-0XXX_ses-1_T1w.nii.gz
Reason:
_T1w.nii[.gz] files must have exactly three dimensions.
Evidence:
this T1w file does not have exactly three dimensions.

I checked the header of one problematic t1w compared to a valid one with fslinfo. Here are the logs:

Problematic t1w:

data_type INT16
dim1 256
dim2 256
dim3 182
dim4 1
datatype 4
pixdim1 1.000000
pixdim2 1.000000
pixdim3 1.000000
pixdim4 0.009861
cal_max 0.0000
cal_min 0.0000
file_type NIFTI-1+

Valid t1w:

data_type INT16
dim1 512
dim2 512
dim3 120
dim4 1
datatype 4
pixdim1 0.498047
pixdim2 0.498047
pixdim3 1.000000
pixdim4 0.015000
cal_max 0.0000
cal_min 0.0000
file_type NIFTI-1+

We can see that, for both, dim4=1.

Do you have any idea of what is happening ? Any solutions ?

Thanks a lot,

mprl

2

That is weird. Could you share the example files?

3

Try fslhd $IMG | grep dim0. I suspect that you’ll find your valid image has dim0 = 3 while your invalid has some larger value.

4

@ChrisGorgolewski Here is one problematic t1w

@effigies It’s giving this output:

dim0 3
pixdim0 0.000000

for both valid and non-valid mri.

5

Could not replicate with the web and CLI validators. Are you using the latest version?

6

Nevermind. I was indeed using the wrong version of the web validator.
Thanks a lot for your help. It is working now.

1 Like
7

@mprl I’m facing the same issue. Which validator version did you use later?

8

Hi @Praitayini_Kanakaraj

We have made several enhancements to the BIDS Validator in the meantime. Are you using the latest version?

Thank you,
Franklin

9

I’m having the same problem when using fmriprep docker latest. Here’s the error message:

bids-validator@1.4.0

1: [ERR] _T1w.nii[.gz] files must have exactly three dimensions. (code: 95 - T1W_FILE_WITH_TOO_MANY_DIMENSIONS)
./sub-06/anat/sub-06_T1w.nii
Evidence: this T1w file does not have exactly three dimensions.

Here’s an example of the problematic t1w file: t1w

Many thanks for your help.

10

dim0 is 4.

$ fslhd Downloads/sub-06_T1w.nii 
filename       Downloads/sub-06_T1w.nii

sizeof_hdr     348
data_type      INT16
dim0           4
dim1           240
dim2           256
dim3           176
dim4           1
dim5           1
dim6           0
dim7           0
[...]

You can clean this up with fslroi:

fslroi <input> <output> 0 1

Rerunning fslhd:

sizeof_hdr     348
data_type      INT16
dim0           3
dim1           240
dim2           256
dim3           176
dim4           1
dim5           1
dim6           1
dim7           1

Difference:

1c1
< filename       Downloads/sub-06_T1w.nii
---
> filename       sub-06_T1w.nii.gz
5c5
< dim0           4
---
> dim0           3
11,12c11,12
< dim6           0
< dim7           0
---
> dim6           1
> dim7           1
66c66
< descrip        MPRAGE P2_ND
---
> descrip        5.0.11
11

It works fine now, many thanks for your help!

12

Hi,

I’m experiencing the same issue. Is there a workaround that does not rely on FSL?

Thanks,

Sebastian

13

Assuming your file is 4D, but with a singleton fourth dimension, you can use nibabel.func.squeeze_image:

import nibabel as nib

img = nib.load("T1w.nii.gz")
img = nib.funcs.squeeze_image(img)
img.to_filename("T1w.nii.gz")
2 Likes