GE Direct field mapping deltaTE (or TE1/TE2)

[mr-jaemin]

What do you think about adding the delta TE or (TE1, TE2) from GE direct field mapping DICOM to BID JSON?

The delta TE = 1e6 / (2 x pi x VelocityEncodeScale)

where VelocityEncodeScale can be found from DICOM

Example value from dcm_qa_fmap/GE_6_CV_Neuro:
(0019,10e2) DS [-69.077667] # 10, 1 VelocityEncodeScale

Delta TE = 2304 us (TE1 = 4608 us, TE2=6912 us)

I confirmed the TE1/TE2/VelocityEncodeScale on the console as well as the above formula from the code. Any suggestion to save the delta TE or TE2 information in the BIDS sidecar JSON file?

GE_6_CV_Neuro_HH_Adv_fieldmaphz.json

{
	"Modality": "MR",
	"MagneticFieldStrength": 3,
	"ImagingFrequency": 127.738,
	"Manufacturer": "GE",
	"ManufacturersModelName": "SIGNA Architect",
	"InstitutionName": "Bay6 NOT FOR DIAGNOSTIC USE",
	"DeviceSerialNumber": "0000000LXBAY6NET",
	"StationName": "GEHCBAY6",
	"BodyPartExamined": "BRAIN",
	"PatientPosition": "HFS",
	"SoftwareVersions": "27\\LX\\MR Software release:DV27.0_R01_1850.a",
	"MRAcquisitionType": "2D",
	"SeriesDescription": "B0map",
	"ProtocolName": "CV_Neuro_HH_Adv",
	"ScanningSequence": "RM",
	"SequenceVariant": "NONE",
	"ScanOptions": "VB_GEMS",
	"ImageType": ["ORIGINAL", "PRIMARY", "OTHER", "REAL", "FIELDMAPHZ"],
	"RawImage": false,
	"SeriesNumber": 6,
	"AcquisitionNumber": 1,
	"Units": "Hz",
	"SliceThickness": 5,
	"SpacingBetweenSlices": 9,
	"SAR": 0.0122642,
	"EchoTime": 0.004608,
	"RepetitionTime": 0.2,
	"FlipAngle": 20,
	"CoilString": "48HAP",
	"PercentPhaseFOV": 100,
	"PercentSampling": 100,
	"AcquisitionMatrixPE": 128,
	"ReconMatrixPE": 128,
	"ParallelReductionFactorInPlane": 2,
	"EffectiveEchoSpacing": 0.00114293,
	"TotalReadoutTime": 0.145152,
	"PixelBandwidth": 488.281,
	"PhaseEncodingAxis": "j",
	"ImageOrientationPatientDICOM": [
		1,
		-0,
		0,
		-0,
		1,
		0	],
	"InPlanePhaseEncodingDirectionDICOM": "COL",
	"ConversionSoftware": "dcm2niix",
	"ConversionSoftwareVersion": "v1.0.20211006"
> }

[mr-jaemin]

The delta TE is controlled by the User CV implicitly:

• If Scan Modes: 1=Patient Scanning (CV2 for 2D or CV30 for 3D), B0 Field Mapping Range is set based on Fat/Water Chemical Shift (i.e. 220 Hz for 3T). In this mode, the UI "B0 Field Mapping Range" (CV4 for 2D) will be displayed but will NOT be utilized.
• If Scan Modes: 0=Service Mode, the B0 Field Mapping Range is set by the UI "B0 Field Mapping Range" (CV4 for 2D or CV33 for 3D)
• DICOM private tags store these CV values (CV2, CV2, CV30, CV33):

2D (B0map or B0rf)

3D (3db0map)

[neurolabusc]

@mr-jaemin I think the BIDS fields are would be EchoTime1 and EchoTime2, right?

[mr-jaemin]

Sounds good! EchoTime1 and EchoTime2 to both magnitude and fieldmap JSON files? Please let me know if you want me to do pull request or you could take care of this.

[neurolabusc]

@mr-jaemin GE scanners generate a direct field mapping. So the only required BIDS field for the FIELDMAPHZ image type is Units which dcm2niix correctly populates (as Hz). The EchoTime1 and EchoTime2 are only required for a phase difference map.

So it looks like adding these values is optional. My sense is that they would only obfuscate the raw magnitude, phase, raw and imaginary images, as each of those only has a single echo time associated with the image. It seems like it might be helpful (but not required) to include EchoTime1 and EchoTime2 in the derived FIELDMAPHZ image as it is the derived from two images with different echo times.

I suggest you submit a pull request, I am recovering from shoulder surgery so typing is laborious for me.

[mr-jaemin]

Agreed that it may be confusing to add EchoTime1 and EchoTime2 to the magnitude images of the first echo (i.e single echo). I would add the EchoTime1 and EchoTime2 to the FIELDMAPHZ JSON file only since they would be a useful information, especially for unwrapping.

I would submit a pull request. Hope you recover soon.

[neurolabusc]

Fixed in development branch