FIX in the lifespan protocol / "Grainy" results after MR-FIX?

121 views
Skip to first unread message

Christian Rubbert

unread,
Jan 16, 2020, 7:25:33 AM1/16/20
to hcp-...@humanconnectome.org
Dear HCP-users,

we have set up the HCP Lifespan protocol on our Siemens 3T Prisma
Scanner using the CMRR multiband sequences with prescan normalize
enabled for all sequences.

At the moment, we are implementing the HCP-Pipeline, using the 4.0.1
release from Github. We adapted the settings (below) for
Examples/Scripts/{PreFreeSurferPipelineBatch,GenericfMRIVolumeProcessingPipelineBatch}.sh
from a post from 2015 (Matthew Glasser;
https://www.mail-archive.com/hcp-...@humanconnectome.org/msg01112.html).

Unfortunately, we're struggling a bit with the FIX step (FIX v1.06.12
with Octave 5.1.0). According to the lifespan protocol, we are acquiring
rfMRI in PA and AP. From what we have read, we think we should apply
multi-run FIX with the following settings (according to
Examples/Scripts/IcaFixProcessingBatch.sh):

fMRINames="rfMRI_REST1_PA@rfMRI_REST1_AP"
ConcatNames="rfMRI_REST1_PA_AP"
bandpass=2000
domot=FALSE
TrainingData=HCP_hp2000.RData
FixThreshold=10
DeleteIntermediates=FALSE

Our questions:

1) Is multi-run FIX the correct approach?
2) Is the bandpass=2000 and domot=FALSE setting sensible for the
lifespan protocol?
3) While after a single-run FIX, the according
rfMRI_REST1_{PA,AP}_hp2000_clean.nii.gz actually still look like a brain
scan, the multi-run FIX rfMRI_REST1_PA_AP_hp2000_clean.nii looks
extremely grainy (see attached screenshot). Is that to be expected?

Thank you very much!

Best regards,
Christian Rubbert

--

PreFreeSurferPipelineBatch.sh:

AvgrdcSTRING="TOPUP"
MagnitudeInputName="NONE"
PhaseInputName="NONE"
TE="NONE"
SpinEchoPhaseEncodeNegative="${StudyFolder}/${Subject}/unprocessed/3T/T1w_MPR1/${Subject}_3T_SpinEchoFieldMap_AP.nii.gz"
SpinEchoPhaseEncodePositive="${StudyFolder}/${Subject}/unprocessed/3T/T1w_MPR1/${Subject}_3T_SpinEchoFieldMap_PA.nii.gz"
SEEchoSpacing="0.00058000872333119893" # Calculated from the DICOM
files: echo "1/(16.577999999999999 * 104)" | bc -l
SEUnwarpDir="x"
TopupConfig="${HCPPIPEDIR_Config}/b02b0.cnf"
GEB0InputName="NONE"
T1wTemplate="${HCPPIPEDIR_Templates}/MNI152_T1_0.8mm.nii.gz"
T1wTemplateBrain="${HCPPIPEDIR_Templates}/MNI152_T1_0.8mm_brain.nii.gz"
T1wTemplate2mm="${HCPPIPEDIR_Templates}/MNI152_T1_2mm.nii.gz"
T2wTemplate="${HCPPIPEDIR_Templates}/MNI152_T2_0.8mm.nii.gz"
T2wTemplateBrain="${HCPPIPEDIR_Templates}/MNI152_T2_0.8mm_brain.nii.gz"
T2wTemplate2mm="${HCPPIPEDIR_Templates}/MNI152_T2_2mm.nii.gz"
TemplateMask="${HCPPIPEDIR_Templates}/MNI152_T1_0.8mm_brain_mask.nii.gz"
Template2mmMask="${HCPPIPEDIR_Templates}/MNI152_T1_2mm_brain_mask_dil.nii.gz"
T1wSampleSpacing="0.0000071"
T2wSampleSpacing="0.0000021"
UnwarpDir="z"
BrainSize="150"
FNIRTConfig="${HCPPIPEDIR_Config}/T1_2_MNI152_2mm.cnf"
GradientDistortionCoeffs="${HCPPIPEDIR_Config}/coeff.grad" # Copied from
our scanner

GenericfMRIVolumeProcessingPipelineBatch.sh:

TaskList=""
TaskList+=" rfMRI_REST1_PA" #Include space as first character
TaskList+=" rfMRI_REST1_AP"
fMRITimeSeries="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_${fMRIName}.nii.gz"
fMRISBRef="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_${fMRIName}_SBRef.nii.gz"
EchoSpacing="0.00058000872333119893" # Calculated from the DICOM files:
echo "1/(16.577999999999999 * 104)" | bc -l
DistortionCorrection="TOPUP"
BiasCorrection="SEBASED"
SpinEchoPhaseEncodeNegative="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_SpinEchoFieldMap_AP.nii.gz"
SpinEchoPhaseEncodePositive="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_SpinEchoFieldMap_PA.nii.gz"
TopUpConfig="${HCPPIPEDIR_Config}/b02b0.cnf"
MagnitudeInputName="NONE"
PhaseInputName="NONE"
DeltaTE="NONE"
GEB0InputName="NONE"
FinalFMRIResolution="2"
GradientDistortionCoeffs="${HCPPIPEDIR_Config}/coeff.grad" # Copied from
our scanner
MCType="MCFLIRT"
rfMRI_REST1_PA_AP_hp2000_clean.png

Glasser, Matthew

unread,
Jan 16, 2020, 9:29:27 AM1/16/20
to hcp-...@humanconnectome.org
I would use hp=0 because it is faster and domot=FALSE is fine. MR+FIX is fine. After single run FIX, the mean fMRI image is retained. After multi-run FIX in the concatenated data, the mean image is not retained (but for single run analysis like task fMRI after multi-run FIX, it is retained).

For PreFreeSurfer ${ SEUnwarpDir}=y because you are using AP/PA.

Thanks for posting your calls to the pipelines with your question. It really makes it much easier to help you.

Matt.
--
You received this message because you are subscribed to the Google Groups "HCP-Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hcp-users+...@humanconnectome.org.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/9d7767ccb1442a498283ab02cbcd6b56%40uni-duesseldorf.de.



________________________________
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.

Harms, Michael

unread,
Jan 16, 2020, 10:47:54 AM1/16/20
to hcp-...@humanconnectome.org

@Matt: Wouldn't you suggest they also use "HCP_Style_Single_Multirun_Dedrift.RData" as the TrainingData?

--
Michael Harms, Ph.D.

-----------------------------------------------------------

Associate Professor of Psychiatry
Washington University School of Medicine
Department of Psychiatry, Box 8134
660 South Euclid Ave. Tel: 314-747-6173
St. Louis, MO 63110 Email: mha...@wustl.edu
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/B480ACFD-00BF-4FF5-A55D-689ADBC12BBC%40wustl.edu.

Glasser, Matthew

unread,
Jan 16, 2020, 10:48:25 AM1/16/20
to hcp-...@humanconnectome.org

Christian Rubbert

unread,
Jan 16, 2020, 1:56:05 PM1/16/20
to hcp-...@humanconnectome.org
Hey,

thank you very much for pointing out the settings we missed!

> MR+FIX is fine. After single run FIX, the mean fMRI image is
> retained. After multi-run FIX in the concatenated data, the mean
> image is not retained (but for single run analysis like task fMRI
> after multi-run FIX, it is retained).

I'm not sure I understand this bit. Does the resting state data after
the MR-FIX pipeline look that way, because it is demeaned and thus
represents only the signal change in each voxel? Does having MR+FIX
change anything in regard to further analyses compared to single-run FIX
where the mean EPI signal is preserved?

Best regards,
Christian

Glasser, Matthew

unread,
Jan 16, 2020, 2:17:53 PM1/16/20
to hcp-...@humanconnectome.org
For MR+FIX, the concatenated data is demeaned. The data split back out into individual runs is not.

Matt.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/0db63ef32e53758592c3201c32181b1a%40uni-duesseldorf.de.

Christian Rubbert

unread,
Feb 2, 2020, 2:48:18 PM2/2/20
to hcp-...@humanconnectome.org
Hey,

we ran the pipeline with MR+FIX on two non-healthy subjects we acquired
in our pilot phase (full settings below). When reviewing the
classification in the Connectome Workbench, we are a bit confused by the
visible break likely due to concatenation of the time series (see
attached screenshots for examples). Is that to be expected, or are we
missing something? Overall, there are only about 10 timeseries
classified as signal out of ~100.

Thanks!

Best regards,
Christian


HCPpipelines-4.0.1 settings:


PreFreeSurferPipelineBatch.sh

AvgrdcSTRING="TOPUP"
MagnitudeInputName="NONE"
PhaseInputName="NONE"
TE="NONE"
SpinEchoPhaseEncodeNegative="${StudyFolder}/${Subject}/unprocessed/3T/T1w_MPR1/${Subject}_3T_SpinEchoFieldMap_AP.nii.gz"
SpinEchoPhaseEncodePositive="${StudyFolder}/${Subject}/unprocessed/3T/T1w_MPR1/${Subject}_3T_SpinEchoFieldMap_PA.nii.gz"
SEEchoSpacing="0.00058000872333119893"
SEUnwarpDir="y"
TopupConfig="${HCPPIPEDIR_Config}/b02b0.cnf"
GEB0InputName="NONE"
T1wTemplate="${HCPPIPEDIR_Templates}/MNI152_T1_0.8mm.nii.gz"
T1wTemplateBrain="${HCPPIPEDIR_Templates}/MNI152_T1_0.8mm_brain.nii.gz"
T1wTemplate2mm="${HCPPIPEDIR_Templates}/MNI152_T1_2mm.nii.gz"
T2wTemplate="${HCPPIPEDIR_Templates}/MNI152_T2_0.8mm.nii.gz"
T2wTemplateBrain="${HCPPIPEDIR_Templates}/MNI152_T2_0.8mm_brain.nii.gz"
T2wTemplate2mm="${HCPPIPEDIR_Templates}/MNI152_T2_2mm.nii.gz"
TemplateMask="${HCPPIPEDIR_Templates}/MNI152_T1_0.8mm_brain_mask.nii.gz"
Template2mmMask="${HCPPIPEDIR_Templates}/MNI152_T1_2mm_brain_mask_dil.nii.gz"
T1wSampleSpacing="0.0000071"
T2wSampleSpacing="0.0000021"
UnwarpDir="z"
BrainSize="150"
FNIRTConfig="${HCPPIPEDIR_Config}/T1_2_MNI152_2mm.cnf"
GradientDistortionCoeffs="${HCPPIPEDIR_Config}/coeff.grad"


FreeSurferPipelineBatch.sh

T1wImage="${StudyFolder}/${Subject}/T1w/T1w_acpc_dc_restore.nii.gz" #T1w
FreeSurfer Input (Full Resolution)
T1wImageBrain="${StudyFolder}/${Subject}/T1w/T1w_acpc_dc_restore_brain.nii.gz"
#T1w FreeSurfer Input (Full Resolution)
T2wImage="${StudyFolder}/${Subject}/T1w/T2w_acpc_dc_restore.nii.gz" #T2w
FreeSurfer Input (Full Resolution)


PostFreeSurferPipelineBatch.sh

SurfaceAtlasDIR="${HCPPIPEDIR_Templates}/standard_mesh_atlases"
GrayordinatesSpaceDIR="${HCPPIPEDIR_Templates}/91282_Greyordinates"
GrayordinatesResolutions="2" #Usually 2mm, if multiple delimit with @,
must already exist in templates dir
HighResMesh="164" #Usually 164k vertices
LowResMeshes="32" #Usually 32k vertices, if multiple delimit with @,
must already exist in templates dir
SubcorticalGrayLabels="${HCPPIPEDIR_Config}/FreeSurferSubcorticalLabelTableLut.txt"
FreeSurferLabels="${HCPPIPEDIR_Config}/FreeSurferAllLut.txt"
ReferenceMyelinMaps="${HCPPIPEDIR_Templates}/standard_mesh_atlases/Conte69.MyelinMap_BC.164k_fs_LR.dscalar.nii"
RegName="MSMSulc" #MSMSulc is recommended, if binary is not available
use FS (FreeSurfer)


GenericfMRIVolumeProcessingPipelineBatch.sh

TaskList=""
TaskList+=" rfMRI_REST1_PA" #Include space as first character
TaskList+=" rfMRI_REST1_AP"
UnwarpDir="y"
UnwarpDir="y-"
UnwarpDir="x"
UnwarpDir="x-"
fMRITimeSeries="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_${fMRIName}.nii.gz"
fMRISBRef="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_${fMRIName}_SBRef.nii.gz"
EchoSpacing="0.00058000872333119893"
DistortionCorrection="TOPUP"
BiasCorrection="SEBASED"
SpinEchoPhaseEncodeNegative="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_SpinEchoFieldMap_AP.nii.gz"
SpinEchoPhaseEncodePositive="${StudyFolder}/${Subject}/unprocessed/3T/${fMRIName}/${Subject}_3T_SpinEchoFieldMap_PA.nii.gz"
TopUpConfig="${HCPPIPEDIR_Config}/b02b0.cnf"
MagnitudeInputName="NONE" #Expects 4D Magnitude volume with two 3D
volumes (differing echo times)
PhaseInputName="NONE" #Expects a 3D Phase difference volume (Siemen's
style)
DeltaTE="NONE" #2.46ms for 3T, 1.02ms for 7T
GEB0InputName="NONE"
FinalFMRIResolution="2"
GradientDistortionCoeffs="${HCPPIPEDIR_Config}/coeff.grad"
MCType="MCFLIRT"


GenericfMRISurfaceProcessingPipelineBatch.sh

Tasklist=""
Tasklist="${Tasklist} rfMRI_REST1_PA"
Tasklist="${Tasklist} rfMRI_REST1_AP"
LowResMesh="32" #Needs to match what is in PostFreeSurfer, 32 is on
average 2mm spacing between the vertices on the midthickness
FinalfMRIResolution="2" #Needs to match what is in fMRIVolume, i.e. 2mm
for 3T HCP data and 1.6mm for 7T HCP data
SmoothingFWHM="2" #Recommended to be roughly the grayordinates spacing,
i.e 2mm on HCP data
GrayordinatesResolution="2" #Needs to match what is in PostFreeSurfer.
2mm gives the HCP standard grayordinates space with 91282 grayordinates.
Can be different from the FinalfMRIResolution (e.g. in the case of HCP
7T data at 1.6mm)
RegName="MSMSulc" #MSMSulc is recommended, if binary is not available
use FS (FreeSurfer)


IcaFixProcessingBatch.sh

fMRINames="rfMRI_REST1_PA@rfMRI_REST1_AP"
ConcatNames="rfMRI_REST1_PA_AP" ## Use space (or @) to separate
concatenation groups
bandpass=0
domot=FALSE # For MR-FIX?
TrainingData=HCP_Style_Single_Multirun_Dedrift.RData
FixThreshold=10
DeleteIntermediates=FALSE


PostFixBatch.sh

fMRINames="rfMRI_REST1_PA_AP"
HighPass="0"
ReUseHighPass="YES" #Use YES if running on output from multi-run FIX,
otherwise use NO
DualScene=${HCPPIPEDIR}/ICAFIX/PostFixScenes/ICA_Classification_DualScreenTemplate.scene
SingleScene=${HCPPIPEDIR}/ICAFIX/PostFixScenes/ICA_Classification_SingleScreenTemplate.scene
MatlabMode=2 #Mode=0 compiled Matlab, Mode=1 interpreted Matlab, Mode=2
octave


MSMAllPipelineBatch.sh

fMRINames=""
mrfixNames="rfMRI_REST1_PA@rfMRI_REST1_AP"
mrfixConcatName="rfMRI_REST1_PA_AP"
mrfixNamesToUse="rfMRI_REST1_PA@rfMRI_REST1_AP"
OutfMRIName="rfMRI_REST_CONCAT"
HighPass="0"
fMRIProcSTRING="_Atlas_hp0_clean"
MSMAllTemplates="${HCPPIPEDIR}/global/templates/MSMAll"
RegName="MSMAll_InitalReg"
HighResMesh="164"
LowResMesh="32"
InRegName="MSMSulc"
MatlabMode="2" #Mode=0 compiled Matlab, Mode=1 interpreted Matlab,
Mode=2 Octave


DeDriftAndResamplePipelineBatch.sh

HighResMesh="164"
LowResMesh="32"
RegName="MSMAll_InitalReg_2_d40_WRN"
DeDriftRegFiles="${HCPPIPEDIR}/global/templates/MSMAll/DeDriftingGroup.L.sphere.DeDriftMSMAll.164k_fs_LR.surf.gii@${HCPPIPEDIR}/global/templates/MSMAll/DeDriftingGroup.R.sphere.DeDriftMSMAll.164k_fs_LR.surf.gii"
ConcatRegName="MSMAll_Test"
Maps="sulc curvature corrThickness thickness"
MyelinMaps="MyelinMap SmoothedMyelinMap" #No _BC, this will be reapplied
MRFixConcatNames="rfMRI_REST1_PA_AP"
MRFixNames="rfMRI_REST1_PA@rfMRI_REST1_AP"
dontFixNames="NONE"
SmoothingFWHM="2" #Should equal previous grayordinates smoothing
(because we are resampling from unsmoothed native mesh timeseries)
HighPass="0"
MotionRegression=FALSE
MatlabMode="2" #Mode=0 compiled Matlab, Mode=1 interpreted Matlab,
Mode=2 octave
Unkown-1.png
Unknown-54.png

Glasser, Matthew

unread,
Feb 2, 2020, 6:45:19 PM2/2/20
to hcp-...@humanconnectome.org
Those both look like noise. What is going on with these settings:

GenericfMRIVolumeProcessingPipelineBatch.sh

TaskList=""
TaskList+=" rfMRI_REST1_PA" #Include space as first character
TaskList+=" rfMRI_REST1_AP"
UnwarpDir="y"
UnwarpDir="y-"
UnwarpDir="x"
UnwarpDir="x-"

AP/PA should be y/y- or y-/y not x.

Matt.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/64353988daca322b8fad11172fc2a2a3%40uni-duesseldorf.de.

Christian Rubbert

unread,
Feb 3, 2020, 6:04:42 AM2/3/20
to hcp-...@humanconnectome.org
Dear Matt,

you are right, we have not seen those effects in timeseries classified
as signal. I used a tiny script to generate these config snippets and
overlooked the UnwarpDir. That's

case ${PhaseEncodingDir} in
"PA")
UnwarpDir="y"
;;
"AP")
UnwarpDir="y-"
;;
"RL")
UnwarpDir="x"
;;
"LR")
UnwarpDir="x-"
;;
*)
echo "${SCRIPT_NAME}: Unrecognized Phase Encoding Direction:
${PhaseEncodingDir}"
exit 1
esac

I have confirmed in my logs, that PA/AP are detected properly. Sorry
about that.

Can you comment on the fact, that only about 10 timeseries were
identified as signal out of roughly 100 in each of our two non-healthy
subjects? Shouldn't we expect more networks being identified? There were
no apparent mis-classifications.

Thank you!

Best regards,
Christian

Steve Smith

unread,
Feb 3, 2020, 6:07:52 AM2/3/20
to hcp-...@humanconnectome.org
Hi

On 3 Feb 2020, at 11:03, Christian Rubbert <chru...@hhu.de> wrote:

Dear Matt,

you are right, we have not seen those effects in timeseries classified as signal. I used a tiny script to generate these config snippets and overlooked the UnwarpDir. That's

case ${PhaseEncodingDir} in
 "PA")
       UnwarpDir="y"
       ;;
 "AP")
       UnwarpDir="y-"
       ;;
 "RL")
       UnwarpDir="x"
       ;;
 "LR")
       UnwarpDir="x-"
       ;;
 *)
       echo "${SCRIPT_NAME}: Unrecognized Phase Encoding Direction: ${PhaseEncodingDir}"
       exit 1
esac

I have confirmed in my logs, that PA/AP are detected properly. Sorry about that.

Can you comment on the fact, that only about 10 timeseries were identified as signal out of roughly 100 in each of our two non-healthy subjects? Shouldn't we expect more networks being identified? There were no apparent mis-classifications.


That probably sounds about right.  Note that only finding 10 at single-subject level does not mean that you can't find (and work with) a larger number once you do (eg) group-level ICA.

Cheers







---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Head of Analysis,  WIN (FMRIB) Oxford

FMRIB, JR Hospital, Headington, Oxford  OX3 9DU, UK
+44 (0) 1865 610470
st...@fmrib.ox.ac.uk    http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------



Glasser, Matthew

unread,
Feb 3, 2020, 9:44:06 AM2/3/20
to hcp-...@humanconnectome.org

How many minutes total fMRI is this?

 

Matt.

Christian Rubbert

unread,
Feb 3, 2020, 1:39:01 PM2/3/20
to hcp-...@humanconnectome.org
Dear Matt,
Dear Steve,

it's the latest HCP lifespan protocol
(HCP_VE11C_CMRRmb_NoPCASL_SiemensT1T2_2019.01.14), unmodified, except
for Prescan Normalize enabled for _all_ sequences. The rs-fMRI should be
6 minutes 40 seconds for each AP and PA.

Thank you for clarifying the potential for group ICA, that had worried
us.

Best,
Christian

Am 2020-02-03 15:44, schrieb Glasser, Matthew:
> How many minutes total fMRI is this?
>
> Matt.
>
> From: Steve Smith <st...@fmrib.ox.ac.uk>
> Reply-To: "hcp-...@humanconnectome.org"
> <hcp-...@humanconnectome.org>
> Date: Monday, February 3, 2020 at 5:07 AM
> To: "hcp-...@humanconnectome.org" <hcp-...@humanconnectome.org>
> Subject: Re: [hcp-users] FIX in the lifespan protocol / "Grainy"
> results after MR-FIX?
>
> Hi
>
>
> On 3 Feb 2020, at 11:03, Christian Rubbert
> <mha...@wustl.edu<mailto:mha...@wustl.edu>> wrote:
> > >
> > >
> > > @Matt: Wouldn't you suggest they also use
> > > "HCP_Style_Single_Multirun_Dedrift.RData" as the
> TrainingData?
> > >
> > > --
> > > Michael Harms, Ph.D.
> > >
> > >
> -----------------------------------------------------------
> > >
> > > Associate Professor of Psychiatry
> > > Washington University School of Medicine
> > > Department of Psychiatry, Box 8134
> > > 660 South Euclid Ave. Tel:
> > 314-747-6173
> > > St. Louis, MO 63110 Email:
> > > mha...@wustl.edu<mailto:mha...@wustl.edu>
> > >
> > > On 1/16/20, 8:29 AM, "Glasser, Matthew"
> <glas...@wustl.edu<mailto:glas...@wustl.edu>>
> > wrote:
> > >
> > > I would use hp=0 because it is faster and domot=FALSE is
> > fine.
> > > MR+FIX is fine. After single run FIX, the mean fMRI image
> is
> > > retained. After multi-run FIX in the concatenated data, the
> mean
> > > image is not retained (but for single run analysis like task
> fMRI
> > > after multi-run FIX, it is retained).
> > >
> > > For PreFreeSurfer ${ SEUnwarpDir}=y because you are
> using
> > AP/PA.
> > >
> > > Thanks for posting your calls to the pipelines with your
> > question.
> > > It really makes it much easier to help you.
> > >
> > > Matt.
> > >
> > > On 1/16/20, 6:24 AM, "Christian Rubbert"
> <chru...@hhu.de<mailto:chru...@hhu.de>>
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> > To view this discussion on the web visit
> >
> https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/0db63ef32e53758592c3201c32181b1a%40uni-duesseldorf.de.
> >
> >
> >
> > ________________________________
> > The materials in this message are private and may contain
> Protected
> > Healthcare Information or other information of a sensitive nature.
> If
> > you are not the intended recipient, be advised that any
> unauthorized
> > use, disclosure, copying or the taking of any action in reliance
> on
> > the contents of this information is strictly prohibited. If you
> have
> > received this email in error, please immediately notify the sender
> via
> > telephone or return mail.
> --
> You received this message because you are subscribed to the Google
> Groups "HCP-Users" group.
> To unsubscribe from this group and stop receiving emails from it,
> send an email to
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> To view this discussion on the web visit
> https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/64353988daca322b8fad11172fc2a2a3%40uni-duesseldorf.de.
> ________________________________
> The materials in this message are private and may contain Protected
> Healthcare Information or other information of a sensitive nature. If
> you are not the intended recipient, be advised that any unauthorized
> use, disclosure, copying or the taking of any action in reliance on
> the contents of this information is strictly prohibited. If you have
> received this email in error, please immediately notify the sender via
> telephone or return mail.
>
> --
> You received this message because you are subscribed to the Google
> Groups "HCP-Users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> To view this discussion on the web visit
> https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/78700c3512905a7c39cb53f6e4183938%40uni-duesseldorf.de.
>
>
> ---------------------------------------------------------------------------
> Stephen M. Smith, Professor of Biomedical Engineering
> Head of Analysis, WIN (FMRIB) Oxford
>
> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
> +44 (0) 1865 610470
> st...@fmrib.ox.ac.uk<mailto:st...@fmrib.ox.ac.uk>
> http://www.fmrib.ox.ac.uk/~steve
> ---------------------------------------------------------------------------
>
>
> [cid:image0...@01D5DA6E.147B72B0]
>
> --
> You received this message because you are subscribed to the Google
> Groups "HCP-Users" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to
> hcp-users+...@humanconnectome.org<mailto:hcp-users+...@humanconnectome.org>.
> To view this discussion on the web visit
> https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/243491E3-9AD6-4AFE-82FD-20FECA622438%40fmrib.ox.ac.uk<https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/243491E3-9AD6-4AFE-82FD-20FECA622438%40fmrib.ox.ac.uk?utm_medium=email&utm_source=footer>.

Harms, Michael

unread,
Feb 3, 2020, 1:49:19 PM2/3/20
to hcp-...@humanconnectome.org

But, did you collect the full four 6.5 min runs that we are collecting for HCP-LS? Because looking back at your call to MR-FIX, you only provided two runs as input...

Cheers,
-MH

--
Michael Harms, Ph.D.

-----------------------------------------------------------

Associate Professor of Psychiatry
Washington University School of Medicine
Department of Psychiatry, Box 8134
660 South Euclid Ave. Tel: 314-747-6173
St. Louis, MO 63110 Email: mha...@wustl.edu

To unsubscribe from this group and stop receiving emails from it, send an email to hcp-users+...@humanconnectome.org.
To view this discussion on the web visit https://groups.google.com/a/humanconnectome.org/d/msgid/hcp-users/ab09a0375a21ff3001bda7fa8920c087%40uni-duesseldorf.de.

Christian Rubbert

unread,
Feb 3, 2020, 2:37:52 PM2/3/20
to hcp-...@humanconnectome.org
Dear Michael,

you're right - we're acquiring only a single session, i.e. an AP run of
6 minutes 40 seconds and a PA run of 6 minutes 40 seconds, totaling 13
minutes 20 seconds. Unfortunately, we can't have the patients come back
for a second session.

Best,
Christian

Timothy Coalson

unread,
Feb 3, 2020, 3:32:20 PM2/3/20
to hcp-users
Keep in mind that the "signal" ICA components used during FIX are basically only used to prevent reduction/modification of real neural signals that happen to correlate with nuisance/artifact signals.  Fewer "signal" components in FIX just means the weaker neural signals aren't protected from the nuisance/artifact regression as much.

Tim


Timothy Coalson

unread,
Feb 3, 2020, 9:36:22 PM2/3/20
to hcp-users
To be clear, when I said 'fewer "signal" components', I was not talking about misclassification: what I mean is that a dataset with more timepoints collected at the same quality could/should result in more FIX ICA components, such that a few more of them represent neural signal, and thus can represent more of the subject's neural signal.  When properly classified, the dataset that allowed ICA to find more neural signal components should protect somewhat more of the total neural signal from being altered by the regression of nuisance and artifact.

Tim

Reply all
Reply to author
Forward
0 new messages