GRE field mapping to nifti generates the file in a different name

[TamerGezici]

Hello,

I am using Windows and the commandline to do dicom to nifti.

I have a GRE_FIELD_MAPPING series and I am converting it to nifti with this command:

dcm2niix.exe -o .\fmap\ -f test .\GRE_FIELD_MAPPING_0015\

Why is the name of the file test_e2_ph.nii ? Shouldn't it just be "test"?

input:

.json output:

{
	"Modality": "MR",
	"MagneticFieldStrength": 3,
	"ImagingFrequency": 123.238,
	"Manufacturer": "Siemens",
	"ManufacturersModelName": "TrioTim",
	"InstitutionName": "UMRAM //AC6DEA//Ankara",
	"InstitutionalDepartmentName": "Department",
	"InstitutionAddress": "Street StreetNo,City,District,TR,ZIP",
	"DeviceSerialNumber": "35397",
	"StationName": "MRC35397",
	"PatientPosition": "HFS",
	"ProcedureStepDescription": "AUSAF^TAMER",
	"SoftwareVersions": "syngo MR B17",
	"MRAcquisitionType": "2D",
	"SeriesDescription": "gre_field_mapping",
	"ProtocolName": "gre_field_mapping",
	"ScanningSequence": "GR",
	"SequenceVariant": "SP",
	"SequenceName": "*fm2d2r",
	"ImageType": ["ORIGINAL", "PRIMARY", "P", "ND", "PHASE"],
	"SeriesNumber": 15,
	"AcquisitionTime": "12:18:12.287500",
	"AcquisitionNumber": 1,
	"SliceThickness": 3,
	"SpacingBetweenSlices": 3.75,
	"SAR": 0.225768,
	"EchoNumber": 2,
	"EchoTime": 0.00765,
	"RepetitionTime": 0.4,
	"FlipAngle": 60,
	"EchoTime1": 0.00519,
	"EchoTime2": 0.00765,
	"PartialFourier": 1,
	"BaseResolution": 64,
	"ShimSetting": [
		-2208,
		8130,
		2011,
		621,
		81,
		-314,
		-244,
		81	],
	"TxRefAmp": 300.819,
	"PhaseResolution": 1,
	"ReceiveCoilName": "32Ch_Head",
	"CoilString": "C:HEA;HEP",
	"PulseSequenceDetails": "%SiemensSeq%\\gre_field_mapping",
	"CoilCombinationMethod": "Adaptive Combine",
	"MatrixCoilMode": "SENSE",
	"PercentPhaseFOV": 100,
	"PercentSampling": 100,
	"PhaseEncodingSteps": 64,
	"AcquisitionMatrixPE": 64,
	"ReconMatrixPE": 64,
	"PixelBandwidth": 260,
	"DwellTime": 3e-05,
	"PhaseEncodingDirection": "j-",
	"SliceTiming": [
		0.20313,
		0,
		0.21094,
		0.00781,
		0.22656,
		0.02344,
		0.24219,
		0.03906,
		0.25,
		0.04688,
		0.26563,
		0.0625,
		0.27344,
		0.07031,
		0.28906,
		0.08594,
		0.30469,
		0.10156,
		0.3125,
		0.10938,
		0.32813,
		0.125,
		0.33594,
		0.13281,
		0.35156,
		0.14844,
		0.36719,
		0.16406,
		0.375,
		0.17188,
		0.39063,
		0.1875	],
	"ImageOrientationPatientDICOM": [
		1,
		0,
		0,
		0,
		0.995227,
		-0.0975829	],
	"ImageOrientationText": "Tra>Cor(-5.6)",
	"InPlanePhaseEncodingDirectionDICOM": "COL",
	"ConversionSoftware": "dcm2niix",
	"ConversionSoftwareVersion": "v1.0.20211006"
}

[neurolabusc]

This is a feature, not a defect and is describing in the file naming documentation. In brief, dcm2niix will append post-fixes like _ph, _real and _imaginary to avoid name clashes with each other and the magnitude image. Data from a single series can be reconstructed to all of these types, and the BIDS specification requires they be saved as separate files. It is very uncommon to only save the phase maps without other images. Tools like FSL's fugue require different representations as described here. Notes on setting up Siemens and Philips scanners are available.

Likewise, the BIDS specification requires that each echo of a multi-echo sequence be saved as a separate file (e.g. 2nd echo is e2) so dcm2niix's behavior ensures BIDS compliance.

As noted you can use --terse to turn off post-fixes (and deal with naming clashes on your own).