I am using dcm2niix to convert DICOM output from a Siemens 7T scanner to Nifti. This usually works fine with anatomical or diffusion scans, but for functional MRI, I get all measurements from a single fMRI run split into separate 3D .nii files. I am expecting a 4D file for each run. Am I missing something?
Hi @rm_neuroimaging , and welcome to Neurostars!
Could you check that you are using a recent release of dcm2niix?
@rm_neuroimaging is your Siemens 7T scanner running VE* or XA* software? Note that XA has evolved rapidly and has introduced new features, so it is important to use a version of dcm2niix that was released after your scanner software was released. In general for VE* it is smart to export EPI data (DWI, fMRI, ASL) as mosaic. With XA* it is crucial to never export EPI data as mosaic, and using the enhanced DICOM provides more meta data than exporting as classic interoperability DICOM (albeit the gap is shrinking with recent releases of XA).
You can take a look at the XA60 DICOM validation set for an example of dcm2niix conversion for Siemens 7T data running XA60. You can explore the json files created by dcm2niix to detect the version of Siemens software and dcm2niix. For example this file from a fMRI sequence lists:
"Manufacturer": "Siemens",
"ManufacturersModelName": "MAGNETOM Terra.X",
"SoftwareVersions": "syngo MR XA60",
"PulseSequenceDetails": "%CustomerSeq%\\cmrr_mbep2d_bold",
"ConversionSoftwareVersion": "v1.0.20241127"
Of particular importance is the PulseSequenceDetails. In this example, it is the popular CMRR sequences. If you have problems with a Siemens product sequence, you should work with the Siemens Research Collaboration Manager affiliated with your center. If you have issues with a reserch sequence, you should work with the developer of the sequence.