Convert Nifti files to BIDS-Format

Hi,
I am wondering if someone found a way to convert Nifti files into a BIDS-Format. I have a dataset where I only have the Nifti files and no dicom files (GSP-Dataset) and need to convert that into a BIDS structure to process the data with fmriprep.

Does anyone have an idea?

Best,
Stephan

The Supplementary Appendices provide the relevant DICOM header field values. This should allow you to create BIDS-format JSON files.

Thanks a lot for your hint. I found a lot of useful information.
But for the functional JSON files information such as SliceTiming etc. for slice time correction is missing.
Then I just cannot do these steps where the data is missing?

Best,
Stephan

The main text reports that the fMRI data used an interleaved ascending order: BOLD runs consisted of 47 interleaved slices (foot—head; 1, 3, 5 … 45, 47, 2, 4, 6 …, 44, 46), both Table 1 and appendix C reveal the TR is 3.0s so you could derive slice timing. However, appendix C also provides the explicit Siemens private slice timing tag 0019, 1029 tag with values in ms: 0, 1537.5, 65, 1602.5, 127.5 .... Since the BIDS SliceTiming tag expects values in seconds, divide the DICOM values by 1000.

Great, thanks a lot!

I followed your advice and created a BIDS Format for the GSP Dataset and further processed the data with fmriprep (https://fmriprep.org/en/stable/).

Now it seems to be that the GSP times series for the different voxels in each subject is quite similar. I receive mainly values for the correlation from the PCC to all other brain areas in between 0.7 and 1, which is too high.

I used for the following sidecar files for all subjects (BIDS):

anat:
{
“StationName”: “MEDPC”,
“SoftwareVersion”: “syngo MR B17”,
“ScanningSequence”: “EP”,
“SequenceVariant”: “SK”,
“ScanOptions”: “FS”,
“SequenceName”: “epfid2d1_72”,
“DwellTime”: 3.1,
“InPlanePhaseEncodingDirection”: “COL”,

"EchoTime": 0.0030,
"RepetitionTime": 3,
"MagneticFieldStrength": 3,
"FlipAngle": 85,
"ManufacturersModelName": "TrioTim",
"ImageType": [
    "ORIGINAL",
    "PRIMARY",
    "M",
    "ND",
    "MOSAIC"
],
"Manufacturer": "Siemens"

and for func:

funct:
{
“StationName”: “MEDPC”,
“SoftwareVersion”: “syngo MR B17”,
“MagneticFieldStrength”: 3,
“ScanningSequence”: “EP”,
“SequenceVariant”: “SK”,
“ScanOptions”: “FS”,
“SequenceName”: “epfid2d1_72”,
“EchoTime”: 0.03,
“DwellTime”: 3.1,
“FlipAngle”: 85,
“InPlanePhaseEncodingDirection”: “COL”,

"RepetitionTime": 3,
"SliceTiming": [
    0,
    1.5375,
    0.065,
    1.6025,
    0.1275,
    1.665,
    0.1925,
    1.730,
    0.255,
    1.795,
    0.320,
    1.8575,
    0.385,
    1.9225,
    0.4475,
    1.985,
    0.5125,
    2.050,
    0.5775,
    2.115,
    0.640,
    2.1775,
    0.705,
    2.2425,
    0.770,
    2.3075,
    0.8325,
    2.370,
    0.8975,
    2.435,
    0.960,
    2.500,
    1.025,
    2.5625,
    1.090,
    2.6275,
    1.1525,
    2.690,
    1.2175,
    2.755,
    1.2825,
    2.820,
    1.345,
    2.8825,
    1.410,
    2.9475,
    1.4725
],
"ManufacturersModelName": "TrioTim",
"TaskName": "fMRI_resting_state",
"ImageType": [
    "ORIGINAL",
    "PRIMARY",
    "M",
    "ND",
    "MOSAIC"
],
"Manufacturer": "Siemens"

}

Are these values for every subject the same? (Slice-Timing…?)

I have the feeling something might be wrong with that and reason why the correlation values are not correct.

Best,

Stephan

I would contact the authors of the source work. I was not involved in these acquisitions, I just read the information they provided.

Ok, probably you are right.
But maybe you know if the fmriprep preprocessing pipeline does enough steps to preprocess entirely bold data? Because steps such as smooting is missing in fmriprep.

Best,