Add manufacturer name for MR Solutions

[halfSpinDoctor]

Describe the bug

Our Varian/Agilent 4.7T preclinical scanner was recently upgraded to MR Solutions. When I convert DICOM output from the scanner, the manufacturer name shows "Philips" instead of "MR Solutions"

Warning: Found 0018,9087 but manufacturer (0008,0070) unknown: assuming Philips.

JSON output shows incorrect manufacturer:
"Manufacturer": "Philips",

To reproduce

Convert data from MR Solutions system.

Expected behavior

Manufacturer should show MR Solutions

Version

Please report the complete version string:

Chris Rorden's dcm2niiX version v1.0.20220720 Clang14.0.0 ARM (64-bit MacOS)
v1.0.20220720

[neurolabusc]

I think this is a duplicate of issue 642 that is addressed in the development branch. You can test this yourself:

git clone --branch development https://github.com/rordenlab/dcm2niix
cd dcm2niix/console
make
./dcm2niix /path/toDICOMs

[neurolabusc]

@halfSpinDoctor your sequence reports Diffusion b-value Attribute (0018,9087), I would also suggest you confirm that the diffusion b-vectors are correctly extracted. These tend to vary by manufacturer and perhaps they even do things different for quadrupeds. I have not seen any validation datasets, so you may want to validate your equipments behavior.

[halfSpinDoctor]

I think this is a #642

Yes, apologies about this. I just converted on the development branch and works as expected.

your sequence reports Diffusion b-value Attribute (0018,9087), I would also suggest you confirm that the diffusion b-vectors are correctly extracted

thanks, yes it is a diffusion sequence, and the vectors are different that what we had prescribed on the system. Working to validate this.

[halfSpinDoctor]

Quick note, in the conversion in the development branch, the bvecs are all 0's (but the bvals are populated and look correct)

[neurolabusc]

bvec definition is often manufacturer specific. Some vendors specify gradients in world space (Philips, Siemens) and others in Physics coordinate space (GE), while the BIDS/FSL convention is for image space. The DICOM public tag 00189601 is unfortunate as opposite polarities (e.g. [1,0,0] and [-1,0,0]) will create an identical matrix but will have different eddy effects which can be modeled with tools like FSL's Eddy.

[neurolabusc]

@halfSpinDoctor if you want the bvecs populated you will either need to submit a pull request or provide sample datasets (you can share data with my institutional email.

[halfSpinDoctor]

I will work on a sample dataset for you to look at, thank you! (Juggling a few things right now, probably will not happen until after Thanksgiving break).