fMRIPrep on T1w files generated by 3dMPRAGEise (the QTAB dataset)

narunpat · May 23, 2024, 10:58pm

We hope to apply fMRIPrep on the QTAB dataset

QTAB includes MP2RAGE files with 2 different inversion times and 1 UNIT1. The author of this database applied 3dMPRAGEise. on these MP2RAGE files to denoise them and created *UNIT1_unbiased_clean.nii.gz in the derivative folder.

As an example, here is sub-0001_ses-01_UNIT1_unbiased_clean.nii.gz, which looks perfectly fine

fMRIPrep on sub-0001 ran successfully without any error, but the results are strange for t1w. We hope to hear your thoughts on this.

Let me explain what we did on the dataset and showed the fMRIPrep results below

For sub-0001, we then copied the

derivatives/UNIT1_denoised/sub-0001/ses-01/anat/sub-0001_ses-01_UNIT1_unbiased_clean.nii.gz

to

sub-0001//ses-01/anat/

(and also did the same thing for ses-02).

To create a JSON for this ‘t1w’ file, we then copied

sub-0001//ses-01/anat/sub-0001_ses-01_UNIT1_denoised.json

and changed its name to

sub-0001_ses-01_t1w.json.

(Note the UNIT1_denoised images are scans denoised straight from the scanner, but are generally still too noisy)

We just changed two things on its json sidecar:

from

  "ImageComments": "DENOISED_IMAGE_(lambda_=_1)",
  "SeriesDescription": "MP2RAGE_wip900C_VE11C_UNI-DEN",

to

 "ImageComments": "3dMPRAGEise",
 "SeriesDescription": "MP2RAGE_wip900C_VE11C_UNI-DEN_3dMPRAGEise",

Probably unrelated but we also changed a few things to the fmap, so that fMRIPrep can correct func images with fmap. They used pepolar fmap here.

For resting state, we followed this suggestion by copying bold files to the fmap folders:
sub-0001//ses-01/fmap/sub-0001_ses-01_task-rest_dir-AP_bold.nii.gz
and
sub-0001//ses-01/fmap/sub-0001_ses-01_task-rest_dir-PA_bold.nii.gz

We then set the
sub-0001_ses-01_task-rest_dir-AP_bold.json to be intended for PA_bold ,i.e., adding

"IntendedFor":"ses-01/fmap/sub-0001_ses-01_task-rest_dir-PA_bold.nii.gz",
"B0FieldIdentifier": "pepolar_fmap0"

and
sub-0001_ses-01_task-rest_dir-PA_bold.json to be intended for
AP_bold, i.e., adding

"IntendedFor":"ses-01/fmap/sub-0001_ses-01_task-rest_dir-AP_bold.nii.gz",
"B0FieldIdentifier": "pepolar_fmap0".

Below are their json files (Note that we didn’t change anything on the task fmri fmap files. From what we gather, fmriprep should be able to deal with this pepolar fmap.)

Also below are the file structure we have for this subject before running fmriprep version: 23.2.2. (Note we tried with or without ignoring t2w, and the results were similar).

/opt/conda/envs/fmriprep/bin/fmriprep /data /out participant --participant_label 0001 --write-graph --notrack --skip_bids_validation --work-dir /home/william/working_dir --ignore=slicetiming --ignore=t2w --cifti-output --fs-subjects-dir freesurfer

“Brain mask and brain tissue segmentation of the T1w” looks quite strange.

“Spatial normalization of the anatomical T1w reference” looks strange as well

Surface reconstruction looks fine

Alignment of functional and anatomical MRI data (coregistration) looks bad.

Maybe we missed something here?

Files:

sub-0001/
sub-0001//ses-01
sub-0001//ses-01/swi
sub-0001//ses-01/swi/sub-0001_ses-01_swi.json
sub-0001//ses-01/swi/sub-0001_ses-01_minIP.nii.gz
sub-0001//ses-01/swi/sub-0001_ses-01_part-phase_GRE.json
sub-0001//ses-01/swi/sub-0001_ses-01_part-mag_GRE.nii.gz
sub-0001//ses-01/swi/sub-0001_ses-01_swi.nii.gz
sub-0001//ses-01/swi/sub-0001_ses-01_part-phase_GRE.nii.gz
sub-0001//ses-01/swi/sub-0001_ses-01_part-mag_GRE.json
sub-0001//ses-01/swi/sub-0001_ses-01_minIP.json
sub-0001//ses-01/fmap
sub-0001//ses-01/fmap/sub-0001_ses-01_task-rest_dir-AP_bold.nii.gz
sub-0001//ses-01/fmap/sub-0001_ses-01_task-rest_dir-PA_bold.json
sub-0001//ses-01/fmap/sub-0001_ses-01_task-rest_dir-PA_bold.nii.gz
sub-0001//ses-01/fmap/sub-0001_ses-01_task-rest_dir-AP_bold.json
sub-0001//ses-01/func
sub-0001//ses-01/func/sub-0001_ses-01_task-rest_dir-AP_bold.nii.gz
sub-0001//ses-01/func/sub-0001_ses-01_task-rest_dir-PA_bold.json
sub-0001//ses-01/func/sub-0001_ses-01_task-rest_dir-PA_bold.nii.gz
sub-0001//ses-01/func/sub-0001_ses-01_task-rest_dir-AP_bold.json
sub-0001//ses-01/dwi
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-01_dwi.json
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-02_dwi.json
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-01_dwi.json
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-01_dwi.nii.gz
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-01_dwi.bvec
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-02_dwi.nii.gz
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-01_dwi.bval
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-01_dwi.nii.gz
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-02_dwi.bval
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-02_dwi.json
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-01_dwi.bval
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-02_dwi.nii.gz
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-01_dwi.bvec
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-02_dwi.bvec
sub-0001//ses-01/dwi/.goutputstream-7NL2M2
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-AP_run-02_dwi.bval
sub-0001//ses-01/dwi/sub-0001_ses-01_dir-PA_run-02_dwi.bvec
sub-0001//ses-01/anat
sub-0001//ses-01/anat/sub-0001_ses-01_FLAIR.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_UNIT1.json
sub-0001//ses-01/anat/sub-0001_ses-01_T2w_TSE_run-03.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_FLAIR.json
sub-0001//ses-01/anat/sub-0001_ses-01_UNIT1.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_UNIT1_denoised.json
sub-0001//ses-01/anat/sub-0001_ses-01_T2w.json
sub-0001//ses-01/anat/sub-0001_ses-01_inv-2_MP2RAGE_defacemask.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_T2w_TSE_run-02.json
sub-0001//ses-01/anat/sub-0001_ses-01_T2w_TSE_run-01.json
sub-0001//ses-01/anat/sub-0001_ses-01_inv-1_MP2RAGE.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_T2w.nii.gz
sub-0001//ses-01/anat/.goutputstream-HBPGN2
sub-0001//ses-01/anat/sub-0001_ses-01_UNIT1_denoised.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_T2w_TSE_run-03.json
sub-0001//ses-01/anat/sub-0001_ses-01_inv-1_MP2RAGE.json
sub-0001//ses-01/anat/sub-0001_ses-01_T2w_TSE_run-02.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_T2w_TSE_run-01.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_inv-2_MP2RAGE.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_inv-2_MP2RAGE.json
sub-0001//ses-01/anat/sub-0001_ses-01_T1w.nii.gz
sub-0001//ses-01/anat/sub-0001_ses-01_T1w.json
sub-0001//ses-01/perf
sub-0001//ses-01/perf/sub-0001_ses-01_asl.json
sub-0001//ses-01/perf/sub-0001_ses-01_m0scan.json
sub-0001//ses-01/perf/sub-0001_ses-01_m0scan.nii.gz
sub-0001//ses-01/perf/sub-0001_ses-01_aslcontext.tsv
sub-0001//ses-01/perf/sub-0001_ses-01_asl.nii.gz
sub-0001//ses-02
sub-0001//ses-02/swi
sub-0001//ses-02/swi/sub-0001_ses-02_part-phase_GRE.json
sub-0001//ses-02/swi/sub-0001_ses-02_part-phase_GRE.nii.gz
sub-0001//ses-02/swi/sub-0001_ses-02_part-mag_GRE.json
sub-0001//ses-02/swi/sub-0001_ses-02_swi.json
sub-0001//ses-02/swi/sub-0001_ses-02_minIP.json
sub-0001//ses-02/swi/sub-0001_ses-02_part-mag_GRE.nii.gz
sub-0001//ses-02/swi/sub-0001_ses-02_swi.nii.gz
sub-0001//ses-02/swi/sub-0001_ses-02_minIP.nii.gz
sub-0001//ses-02/fmap
sub-0001//ses-02/fmap/sub-0001_ses-02_task-rest_dir-PA_bold.json
sub-0001//ses-02/fmap/sub-0001_ses-02_dir-PA_epi.json
sub-0001//ses-02/fmap/sub-0001_ses-02_dir-PA_epi.nii.gz
sub-0001//ses-02/fmap/sub-0001_ses-02_task-rest_dir-AP_bold.nii.gz
sub-0001//ses-02/fmap/sub-0001_ses-02_task-rest_dir-AP_bold.json
sub-0001//ses-02/fmap/sub-0001_ses-02_dir-AP_epi.nii.gz
sub-0001//ses-02/fmap/sub-0001_ses-02_dir-AP_epi.json
sub-0001//ses-02/fmap/sub-0001_ses-02_task-rest_dir-PA_bold.nii.gz
sub-0001//ses-02/func
sub-0001//ses-02/func/sub-0001_ses-02_task-rest_dir-PA_bold.json
sub-0001//ses-02/func/sub-0001_ses-02_task-partlycloudy_events.json
sub-0001//ses-02/func/sub-0001_ses-02_task-rest_dir-AP_bold.nii.gz
sub-0001//ses-02/func/sub-0001_ses-02_task-partlycloudy_events.tsv
sub-0001//ses-02/func/sub-0001_ses-02_task-rest_dir-AP_bold.json
sub-0001//ses-02/func/sub-0001_ses-02_task-partlycloudy_bold.json
sub-0001//ses-02/func/sub-0001_ses-02_task-partlycloudy_bold.nii.gz
sub-0001//ses-02/func/sub-0001_ses-02_task-rest_dir-PA_bold.nii.gz
sub-0001//ses-02/dwi
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-01_dwi.bvec
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-02_dwi.bvec
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-02_dwi.bval
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-01_dwi.json
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-01_dwi.bval
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-02_dwi.bval
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-02_dwi.bvec
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-01_dwi.nii.gz
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-01_dwi.json
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-01_dwi.nii.gz
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-02_dwi.nii.gz
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-PA_run-02_dwi.json
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-02_dwi.nii.gz
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-01_dwi.bvec
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-02_dwi.json
sub-0001//ses-02/dwi/sub-0001_ses-02_dir-AP_run-01_dwi.bval
sub-0001//ses-02/anat
sub-0001//ses-02/anat/sub-0001_ses-02_T2w_TSE_run-02.json
sub-0001//ses-02/anat/sub-0001_ses-02_FLAIR.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_inv-2_MP2RAGE_defacemask.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_T2w_TSE_run-01.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_inv-1_MP2RAGE.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_UNIT1.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_T1w.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_inv-2_MP2RAGE.json
sub-0001//ses-02/anat/sub-0001_ses-02_inv-1_MP2RAGE.json
sub-0001//ses-02/anat/sub-0001_ses-02_T2w_TSE_run-01.json
sub-0001//ses-02/anat/sub-0001_ses-02_inv-2_MP2RAGE.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_UNIT1_denoised.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_T2w.json
sub-0001//ses-02/anat/sub-0001_ses-02_T1w.json
sub-0001//ses-02/anat/sub-0001_ses-02_UNIT1.json
sub-0001//ses-02/anat/sub-0001_ses-02_UNIT1_denoised.json
sub-0001//ses-02/anat/sub-0001_ses-02_FLAIR.json
sub-0001//ses-02/anat/sub-0001_ses-02_T2w_TSE_run-02.nii.gz
sub-0001//ses-02/anat/sub-0001_ses-02_T2w.nii.gz
sub-0001//ses-02/perf
sub-0001//ses-02/perf/sub-0001_ses-02_asl.json
sub-0001//ses-02/perf/sub-0001_ses-02_asl.nii.gz
sub-0001//ses-02/perf/sub-0001_ses-02_m0scan.json
sub-0001//ses-02/perf/sub-0001_ses-02_m0scan.nii.gz
sub-0001//ses-02/perf/sub-0001_ses-02_aslcontext.tsv

ses-01/anat/sub-0001_ses-01_T1w.json

{
  "AcquisitionMatrixPE": 300,
  "AcquisitionNumber": 1,
  "AcquisitionTime": "11:20:19.000000",
  "BaseResolution": 320,
  "BodyPartExamined": "BRAIN",
  "CoilString": "HC1-7;NC1_2",
  "ConsistencyInfo": "N4_VE11C_LATEST_20160120",
  "ConversionSoftware": "dcm2niix",
  "ConversionSoftwareVersion": "v1.0.20190410  GCC6.3.0",
  "DeviceSerialNumber": "167001",
  "DwellTime": 6.5e-06,
  "EchoTime": 0.00299,
  "ImageComments": "3dMPRAGEise",
  "ImageOrientationPatientDICOM": [-0.0652318, 0.99787, 9.3308e-09, 0.0330828, 0.00216267, -0.99945],
  "ImageType": [
    "DERIVED",
    "PRIMARY",
    "M",
    "UNI",
    "DIS3D",
    "DIS2D"
],
  "ImagingFrequency": 123.229,
  "InPlanePhaseEncodingDirectionDICOM": "ROW",
  "InstitutionAddress": "Cooper_60_St._Lucia_QLD_AU_4072",
  "InstitutionName": "St_Lucia_Campus",
  "InstitutionalDepartmentName": "Gehrmann_Labs_MRI",
  "MRAcquisitionType": "3D",
  "MagneticFieldStrength": 3,
  "Manufacturer": "Siemens",
  "ManufacturersModelName": "Prisma_fit",
  "Modality": "MR",
  "PartialFourier": 0.75,
  "PatientPosition": "HFS",
  "PercentPhaseFOV": 93.75,
  "PhaseEncodingSteps": 226,
  "PhaseResolution": 1,
  "PixelBandwidth": 240,
  "ProcedureStepDescription": "Research_16092_QTAB",
  "ProtocolName": "MP2RAGE_wip900C_VE11C",
  "PulseSequenceDetails": "%CustomerSeq%_tfl_wip900C_VE11C",
  "RawImage": false,
  "ReceiveCoilActiveElements": "HC1-7;NC1,2",
  "ReceiveCoilName": "HeadNeck_64",
  "ReconMatrixPE": 300,
  "RefLinesPE": 24,
  "RepetitionTime": 4,
  "SAR": 0.0695932,
  "ScanOptions": "IR_PFP",
  "ScanningSequence": "GR_IR",
  "SequenceName": "tfl3d1_16ns",
  "SequenceVariant": "SK_SP_MP",
  "SeriesDescription": "MP2RAGE_wip900C_VE11C_UNI-DEN_3dMPRAGEise",
  "SeriesNumber": 8,
  "ShimSetting": [3036, -4179, -850, 100, -62, -225, 52, 12],
  "SliceThickness": 0.8,
  "SoftwareVersions": "syngo_MR_E11",
  "StationName": "mrc_Trio",
  "TxRefAmp": 206.955,
  "WipMemBlock": "WIP_Identifier: WIP#900C"
}

ses-01/fmap/sub-0001_ses-01_task-rest_dir-AP_bold.json

{
  "AcquisitionMatrixPE": 104,
  "AcquisitionNumber": 1,
  "AcquisitionTime": "11:28:16.240000",
  "BandwidthPerPixelPhaseEncode": 17.17,
  "BaseResolution": 104,
  "BodyPartExamined": "BRAIN",
  "ConsistencyInfo": "N4_VE11C_LATEST_20160120",
  "ConversionSoftware": "dcm2niix",
  "ConversionSoftwareVersion": "v1.0.20171215 GCC6.3.0",
  "DerivedVendorReportedEchoSpacing": 0.000560011,
  "DeviceSerialNumber": "167001",
  "DwellTime": 2.1e-06,
  "EchoTime": 0.03,
  "EchoTrainLength": 91,
  "EffectiveEchoSpacing": 0.000560011,
  "FlipAngle": 52,
  "ImageOrientationPatientDICOM": [0.997551, 0.0664439, 0.0218547, -0.0550732, 0.938738, -0.340204],
  "ImageType": [
    "ORIGINAL",
    "PRIMARY",
    "M",
    "ND",
    "MOSAIC"
],
  "InPlanePhaseEncodingDirectionDICOM": "COL",
  "InstitutionAddress": "Cooper_60_St._Lucia_QLD_AU_4072",
  "InstitutionName": "St_Lucia_Campus",
  "InstitutionalDepartmentName": "Gehrmann_Labs_MRI",
  "MRAcquisitionType": "2D",
  "MagneticFieldStrength": 3,
  "Manufacturer": "Siemens",
  "ManufacturersModelName": "Prisma_fit",
  "Modality": "MR",
  "MultibandAccelerationFactor": 6,
  "PartialFourier": 0.875,
  "PatientPosition": "HFS",
  "PercentPhaseFOV": 100,
  "PhaseEncodingDirection": "j-",
  "PhaseEncodingSteps": 91,
  "PhaseResolution": 1,
  "PixelBandwidth": 2290,
  "ProcedureStepDescription": "Research_16092_QTAB",
  "ProtocolName": "ep2d_REST_SMS6_A-P",
  "PulseSequenceDetails": "%SiemensSeq%_ep2d_bold",
  "ReceiveCoilActiveElements": "HC1-7;NC1,2",
  "ReceiveCoilName": "HeadNeck_64",
  "ReconMatrixPE": 104,
  "RepetitionTime": 0.93,
  "SAR": 0.0683576,
  "ScanOptions": "PFP_FS",
  "ScanningSequence": "EP",
  "SequenceName": "_epfid2d1_104",
  "SequenceVariant": "SK",
  "SeriesDescription": "ep2d_REST_SMS6_A-P",
  "SeriesNumber": 9,
  "ShimSetting": [3037, -4137, -746, 577, -94, 157, -69, 54],
  "SliceThickness": 2,
  "SliceTiming": [0.6825, 0, 0.455, 0.075, 0.53, 0.1525, 0.6075, 0.3025, 0.7575, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.075, 0.53, 0.1525, 0.6075, 0.3025, 0.7575, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.075, 0.53, 0.1525, 0.6075, 0.3025, 0.7575, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.075, 0.53, 0.1525, 0.6075, 0.3025, 0.7575, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.075, 0.53, 0.1525, 0.6075, 0.3025, 0.7575, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.075, 0.53, 0.1525, 0.6075, 0.3025, 0.7575, 0.38, 0.835, 0.2275],
  "SoftwareVersions": "syngo_MR_E11",
  "SpacingBetweenSlices": 2,
  "StationName": "mrc_Trio",
  "TaskName": "rest",
  "TotalReadoutTime": 0.0576811,
  "TxRefAmp": 206.955,
  "IntendedFor":"ses-01/fmap/sub-0001_ses-01_task-rest_dir-PA_bold.nii.gz",
  "B0FieldIdentifier": "pepolar_fmap0"
}

ses-01/fmap/sub-0001_ses-01_task-rest_dir-PA_bold.json

{
  "AcquisitionMatrixPE": 104,
  "AcquisitionNumber": 1,
  "AcquisitionTime": "11:33:56.935000",
  "BandwidthPerPixelPhaseEncode": 17.17,
  "BaseResolution": 104,
  "BodyPartExamined": "BRAIN",
  "ConsistencyInfo": "N4_VE11C_LATEST_20160120",
  "ConversionSoftware": "dcm2niix",
  "ConversionSoftwareVersion": "v1.0.20171215 GCC6.3.0",
  "DerivedVendorReportedEchoSpacing": 0.000560011,
  "DeviceSerialNumber": "167001",
  "DwellTime": 2.1e-06,
  "EchoTime": 0.03,
  "EchoTrainLength": 91,
  "EffectiveEchoSpacing": 0.000560011,
  "FlipAngle": 52,
  "ImageOrientationPatientDICOM": [0.997551, 0.0664439, 0.0218547, -0.0550732, 0.938738, -0.340204],
  "ImageType": [
    "ORIGINAL",
    "PRIMARY",
    "M",
    "ND",
    "MOSAIC"
],
  "InPlanePhaseEncodingDirectionDICOM": "COL",
  "InstitutionAddress": "Cooper_60_St._Lucia_QLD_AU_4072",
  "InstitutionName": "St_Lucia_Campus",
  "InstitutionalDepartmentName": "Gehrmann_Labs_MRI",
  "MRAcquisitionType": "2D",
  "MagneticFieldStrength": 3,
  "Manufacturer": "Siemens",
  "ManufacturersModelName": "Prisma_fit",
  "Modality": "MR",
  "MultibandAccelerationFactor": 6,
  "PartialFourier": 0.875,
  "PatientPosition": "HFS",
  "PercentPhaseFOV": 100,
  "PhaseEncodingDirection": "j",
  "PhaseEncodingSteps": 91,
  "PhaseResolution": 1,
  "PixelBandwidth": 2290,
  "ProcedureStepDescription": "Research_16092_QTAB",
  "ProtocolName": "ep2d_REST_SMS6_P-A",
  "PulseSequenceDetails": "%SiemensSeq%_ep2d_bold",
  "ReceiveCoilActiveElements": "HC1-7;NC1,2",
  "ReceiveCoilName": "HeadNeck_64",
  "ReconMatrixPE": 104,
  "RepetitionTime": 0.93,
  "SAR": 0.0683576,
  "ScanOptions": "PFP_FS",
  "ScanningSequence": "EP",
  "SequenceName": "_epfid2d1_104",
  "SequenceVariant": "SK",
  "SeriesDescription": "ep2d_REST_SMS6_P-A",
  "SeriesNumber": 10,
  "ShimSetting": [3037, -4137, -746, 577, -94, 157, -69, 54],
  "SliceThickness": 2,
  "SliceTiming": [0.6825, 0, 0.455, 0.0775, 0.5325, 0.1525, 0.6075, 0.305, 0.76, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.0775, 0.5325, 0.1525, 0.6075, 0.305, 0.76, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.0775, 0.5325, 0.1525, 0.6075, 0.305, 0.76, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.0775, 0.5325, 0.1525, 0.6075, 0.305, 0.76, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.0775, 0.5325, 0.1525, 0.6075, 0.305, 0.76, 0.38, 0.835, 0.2275, 0.6825, 0, 0.455, 0.0775, 0.5325, 0.1525, 0.6075, 0.305, 0.76, 0.38, 0.835, 0.2275],
  "SoftwareVersions": "syngo_MR_E11",
  "SpacingBetweenSlices": 2,
  "StationName": "mrc_Trio",
  "TaskName": "rest",
  "TotalReadoutTime": 0.0576811,
  "TxRefAmp": 206.955,
  "IntendedFor":"ses-01/fmap/sub-0001_ses-01_task-rest_dir-AP_bold.nii.gz",
  "B0FieldIdentifier": "pepolar_fmap0"
}

@tsalo I tagged you here since I saw your post on QTAB. I would hope to hear your thoughts on this.
Thanks so much.
Narun

narunpat · June 7, 2024, 10:39am

Just a quick update here. I tried a quick and dirty afniproc.py (i.e., without freesurfer, see below), and the results look fine (see attached below). I was wondering what might have gone wrong with fmriprep.

#!/usr/bin/env tcsh

set subj = sub-0001
set ses = ses-01

set top_dir = /media/hcs-sci-psy-narun/QTAB/copy/ds004146-download/${subj}/${ses}
set anat_dir = $top_dir/anat
set func_dir = $top_dir/func

#run afni_proc.py to create a single subject preprocessing script
afni_proc.py                                                         \
    -subj_id                  ${subj}						                     	 \
    -script                   proc.${subj}.${ses}                        \
    -scr_overwrite                                                   \
    -blocks                   align tlrc volreg mask blur     \
                              scale regress                          \
    -copy_anat                ${anat_dir}/${subj}_${ses}_T1w.nii.gz        \
    -anat_has_skull           yes                                    \
    -dsets                    ${func_dir}/${subj}_${ses}_task-rest_dir-AP_bold.nii.gz \
                              ${func_dir}/${subj}_${ses}_task-rest_dir-PA_bold.nii.gz \
    -blip_forward_dset        ${func_dir}/${subj}_${ses}_task-rest_dir-AP_bold.nii.gz \
    -blip_reverse_dset        ${func_dir}/${subj}_${ses}_task-rest_dir-PA_bold.nii.gz \
    -tcat_remove_first_trs    0                                      \
    -radial_correlate         yes                                    \
    -align_opts_aea           -cost lpc+ZZ                           \
                              -giant_move                            \
                              -check_flip                            \
    -tlrc_base                MNI152_2009_template_SSW.nii.gz        \
    -tlrc_NL_warp                                                    \
    -volreg_align_to          MIN_OUTLIER                            \
    -volreg_align_e2a                                                \
    -volreg_tlrc_warp                                                \
    -mask_epi_anat            yes                                    \
    -blur_size                5                                      \
    -regress_motion_per_run                                          \
    -regress_censor_motion    0.3                                    \
    -regress_censor_outliers  0.05                                   \
    -regress_est_blur_errts                                          \
    -regress_run_clustsim     no                                     \
    -html_review_style        pythonic

bashirms · June 21, 2024, 1:46pm

Hello! Did you wind up figuring out what went wrong with fmriprep here? I am currently having the same issue with similar tissue segmentation and spatial normalization outputs.

ptaylor · June 21, 2024, 8:52pm

Hi, @narunpat -

Glad the afni_proc.py command worked.

One suggestion: to improve EPI-anatomical alignment, I would try adding this option:

-align_unifize_epi        local

That helps with EPI-anatomical alignment in cases of EPI inhomogeneity, which are present here.

In fact, I would generally use it on all human FMRI processing, because it generally seems to be generally helpful in EPI-anatomical alignment. (In nonhuman FMRI processing, there is more non-brain tissue in the EPI FOV, which complicates things, and so I wouldn’t add this option at present.)

–pt

narunpat · August 3, 2024, 10:05am

Thank you so much @ptaylor. Appreciate your suggestion. We ended up fixing the fmriprep in this case, but I would love to hear from you, for educational purposes, how could you tell if there is “EPI inhomogeneity” based on the AFNI html report?

@bashirms sorry for the late reply. We updated the fmirprep from 23.2.2 to 24.0.1, and somehow the problem went away. We also changed a bit of the bids fmap structure (e.g., using B0FieldIdentifier as opposed to IntendedFor etc.), but this might not be related to the issue of T1w shown here.

ptaylor · August 3, 2024, 11:52pm

@narunpat :

Re. homogeneity/inhomogeneity:

If you think about the average brightness across a particular tissue, or typical brightness level across the brain, this gets at the issue. Inhomogeneous datasets have average-brighter regions, often with what is physically closer to a coil, but this can arise for many reasons. Also, signal intensity can often drop as one goes deeper into the brain, so it might appear relatively darker there. As a consequence of both phenomena, one will often see the relative level of tissue contrast (roughly, how easy it is to differentiate GM from WM from ventricle). In extreme cases, the ability to differentiate tissues can be greatly reduced or removed, making it hard to see sulcal/gyral patterns. It is the contrast that typically drives alignment—differentiating structure means you can try to match it across volumes, which is what cost functions will aim to do.

Consider the two images below—I would say the top row has much more homogeneous EPI than the lower one. In the lower one, the posterior part of the brain glows much more brightly than the middle or frontal. Also, the inner part of the brain appears much darker, even for a particular tissue type. It is almost difficult to tell the ventricle boundaries from WM, which is not the case in the top.

–pt

narunpat · August 4, 2024, 7:59am

thank you so much for your clear explanation!