Siemens Import Errors

Submitted by lm1619 on

Hello Tortoise development team,

We are trying to use the TORTOISE V3.2 software package to postprocess Siemens SE DTI images of ex vivo hearts.

We are having some issues with the data import step.

When trying to import the dicom folder with the ImportDICOM    -i     folder_name command, we get the following warning and error:

Based on Chris Rorden's dcm2niiX version v1.0.20170624
WARNING!! Only ONE b-value in the data!!!   Warning: DTI gradient directions only tested for axial (transverse) acquisitions. Please validate bvec files.

No files to combine...Exiting...
File file_name.nii not a 4D image file.. Exiting

We have also tried to first convert the dicom files to NIfTI format with the dcm2niix converter (v1.0.20200331). It performs the conversion but we get the following warnings:

Warning: interpolated protocol 'SE_1p4mmISO_30dir_5sl_100av_TR2000' may be unsuitable for dwidenoise/mrdegibbs. diffusion_images/MR_245908_00001.dcm

Warning: Check that 2D images are not mirrored.

Despite these warnings, it creates 4 files for each dicom image: .nii, .json, .bval and .bvec. 

When using the ImportNIFTI command as indicated in the web pipeline, we get the same error as with the ImportDICOM :

File file_name.nii not a 4D image file.. Exiting

We would appreciate any advice or help you could offer and would be happy to share our data with you if that could help.

Thank you very much.

Submitted by irfanogl on Wed, 07/08/2020 - 12:13



Unfortunately, our import routines are designed to reject only 3D data.

IF the 4 NIFTI files that dcm2niix generates are in fact the 4 volumes in a single dataset,  you should just concatenate those with for example fslmerge.

Then you can manually concatenate the bvecs/bvals files and use TORTOISE's ImportNIFTI to import your data.


Submitted by lm1619 on Wed, 07/15/2020 - 10:33

In reply to by irfanogl


Thank you very much for your reply. I am afraid we are quite new at using your software and do not fully understand your reply. 

Our dataset in this case is comprised of only 5 2D slices (we started small but our final data sets should have many more slices, maybe 50). Each slice has 6300 dicoms: b=0 (300 averages) + b=150 (30 diffusion directions and 100 averages) + b=750 (30 diffusion directions and 100 averages). These are Siemens dicoms and we used the standard diffusion encoding so we have no external text file with diffusion directions, the b-value and direction info is in the dicoms themselves.

What do you consider a 4D image? And also, you refer to “4 volumes” and we are not sure what those volumes are. What is a “volume” in your software?

We have tried to read the info in the web page and the papers but I am afraid that we still do not fully understand.

Thank you very much in advance.