inconsistent `.json` metadata contents across subjects in `rec-composed` T1w and T2w

[samuellestonge]

I noticed that the contents of the .json sidecar files for the T1w_rec-composed and T2w_rec-composed images vary between subjects in the philadelphia-pediatric dataset.

More specifically:

• Subjects sub-101 to sub-124 (which correspond to the initial data added to neurogitea) have json files contain the following fields:

{
	"ConversionSoftware" ,
	"SeriesNumber" ,
	"SeriesDescription",
	"ImageType",
	"Modality" ,
	"AcquisitionDateTime", 
	"PatientName" ,
	"PatientSex",
	"PatientAge",
	"PatientSize",
	"PatientWeight",
	"PatientPosition",
	"Manufacturer",
	"SoftwareVersion",
	"InstitutionName",
	"InstitutionAddress",
	"DeviceSerialNumber"
}

• Subjects added later (sub-125 to sub-218) have the following fields, consistent with those found in json sidecars for the other contrasts (dwi and MT), and for the acq-top and acq-bottom T1w and T2w files.

{
    "Modality",
    "MagneticFieldStrength",
    "ImagingFrequency",
    "Manufacturer",
    "ManufacturersModelName",
    "InstitutionName",
    "InstitutionalDepartmentName",
    "InstitutionAddress",
    "DeviceSerialNumber",
    "StationName",
    "BodyPartExamined",
    "PatientPosition",
    "ProcedureStepDescription",
    "SoftwareVersions",
    "MRAcquisitionType",
    "SeriesDescription",
    "ProtocolName",
    "ScanningSequence",
    "SequenceVariant",
    "ScanOptions",
    "SequenceName",
    "ImageType",
    "RawImage",
    "SeriesNumber",
    "AcquisitionTime",
    "AcquisitionNumber": 1,
    "SliceThickness",
    "SAR",
    "NumberOfAverages",
    "EchoTime",
    "RepetitionTime",
    "SpoilingState",
    "FlipAngle",
    "AveragesDouble",
    "PartialFourier",
    "BaseResolution",
    "ShimSetting",
    "TxRefAmp",
    "PhaseResolution",
    "PhaseOversampling",
    "ReceiveCoilName",
    "ReceiveCoilActiveElements",
    "PulseSequenceDetails",
    "RefLinesPE",
    "CoilCombinationMethod",
    "ConsistencyInfo",
    "MatrixCoilMode",
    "PercentPhaseFOV",
    "PercentSampling",
    "EchoTrainLength",
    "PhaseEncodingSteps",
    "AcquisitionMatrixPE",
    "ReconMatrixPE",
    "ParallelReductionFactorInPlane",
    "PixelBandwidth",
    "DwellTime",
    "PhaseEncodingDirection",
    "ImageOrientationPatientDICOM",
    "ImageOrientationText",
    "InPlanePhaseEncodingDirectionDICOM",
    "ConversionSoftware",
    "ConversionSoftwareVersion",
    "Dcm2bidsVersion"
}

I am opening this issue to discuss how we should harmonize the metadata for the rec-composed json files across all subjects.

On another note, I would like to add "PatientWeight" and "PatientSize" to the participants.tsv file. However, I was only able to find this information for sub-101 to sub-124 in their rec-composed T1w/T2w .json sidecars. I'm wondering if there's a way we could retrieve this information for all subjects.

[mguaypaq]

Do you have access to the original DICOM files? If so, maybe you could re-convert the older images with a more recent version of dcm2niix (instead of dicm2nii.m 20200911).