Hi Chris,

Hope you are well. I am struggling to convert some clinical data I received from a GE Signa 750. I've ensured I've downloaded the lateset release for dcm2niix to handle some of the quirks of GE and it seems to convert some images but not others. The one image I want (dwi) is failing.

This is the current error message I get: 

Chris Rorden's dcm2niiX version v1.0.20240202  (JP2:OpenJPEG) (JP-LS:CharLS) Clang14.0.3 ARM (64-bit MacOS)

Found 3960 DICOM file(s)

Image Decompression is new: please validate conversions

Segmentation fault: 11

Would you have any suggestions on what to do?

Thanks,

Amy

Does decompressing your DICOMs show errors with other tools? If not, can you convert the decompressed images with dcm2niix? You can use dcmdjpeg or gdcmconv to decompress your images, e.g.

gdcmconv -w in.dcm out.dcm

Hi Chris,

Sorry for the delay. Would I need to decompress each individual .dcm within a folder and then try to run dcm2niix again? 

For example in this folder alone there are ~3959 dicoms. I've just tried running gdcmconc -w command on a single .dcm file and it seems to have worked. Assume I just need to loop through them all to apply this decompression and then try dcm2niix again?

Thanks,

Amy

I decompressed the DICOMs in this folder looping through using gdcmconv but still get a segmentation fault (without the Image Decompression issue).

Chris Rorden's dcm2niiX version v1.0.20240202  (JP2:OpenJPEG) (JP-LS:CharLS) Clang12.0.0 ARM (64-bit MacOS)

Found 3960 DICOM file(s)

Segmentation fault: 11

Thanks,

Amy

Originally posted by Amy Jolly:

Hi Chris,

Hope you are well. I am struggling to convert some clinical data I received from a GE Signa 750. I've ensured I've downloaded the lateset release for dcm2niix to handle some of the quirks of GE and it seems to convert some images but not others. The one image I want (dwi) is failing.

This is the current error message I get: 

Chris Rorden's dcm2niiX version v1.0.20240202  (JP2:OpenJPEG) (JP-LS:CharLS) Clang14.0.3 ARM (64-bit MacOS)

Found 3960 DICOM file(s)

Image Decompression is new: please validate conversions

Segmentation fault: 11

 

Would you have any suggestions on what to do?

 

Thanks,

Amy

Amy

Thanks for the sample dataset. The problem is described here

  https://github.com/rordenlab/dcm2niix/is...

This problem is now fixed in the development branch - and will be included in the next major release. In the mean time, you can compile your own copy of the development branch. As a bonus, if you build with the USE_OPENJPEG option it will convert your rare DICOM compression transfer syntax. So the steps would be:

git clone --branch development https://github.com/rordenlab/dcm2niix.gi...
cd dcm2niix
mkdir build && cd build
cmake -DZLIB_IMPLEMENTATION=Cloudflare -DUSE_JPEGLS=ON -DUSE_OPENJPEG=ON ..
make
cd bin
./dcm2niix /path/to/dicoms

This was a very unusual sequence, with a lot of derived images and RGB colored screenshots. Since the software version was stripped from the DICOMs, you will need to  manually insert the slice timing fields into your EPI data. You may want to check the provenance of these images to see why the public software version tag was removed from your data - that is helpful for audits and to control variability. Beyond these points, the images themselves look terrific - the GE Premier's name is truth in advertising.

Chris,

Thank you so much for your help. I suspect it is because the images are only provided via download from a 3DNet health portal, but I am guessing here!

Best wishes,

Amy

Understood. I would suggest you reach out to the maintainers of the health portal to understand why these DICOMs lack the Software Versions (0018,1020) tag. It is a Type 3 (optional) public tag, so it is not required by a valid DICOM. However, this is not only required to determine slice timing, but is crucial for auditing errors or changes in sequences. I would not consider these images to be of archival qaulity.