Re: [HCP-Users] Slice-timing correction and latency structure of resting-state fMRI data

[Timothy Coalson]

Technically, what should matter more is what frequencies dominate the
correlation in the data of interest - if they are lower enough in frequency
than your sample rate (and therefore slice timing spread), then each
latency should produce reasonable maps.  However, this also means that when
you *compare* two different latencies (you might be comparing versus zero
latency), they should generally be separated by more than those same
frequencies (and therefore, by more than the sample rate), or else you are
oversampling them.

A disclaimer, though: I haven't done this kind of analysis, I'm just
working from the principles I know.

Tim


On Tue, Sep 5, 2017 at 8:48 AM, HINDRIKS, RIKKERT 
wrote:

>
> Dear all,
>
> I am analyzing the latency structure of some of the HCP resting-state fMRI
> data and I want to make sure that this makes sense, given that no
> slice-timing correction has been applied to the data. I would appreciate it
> if someone could confirm that the following reasoning is correct:
>
> Since the entire brain is scanned in 0.78 seconds (one sample), it does no
> make sense to analyze signals latencies < 1 sample, because such small
> latencies will be distorted. In particular, it does not make sense to
>  interpolate the cross-covariance functions as done in one of Mitra's
> papers).
>
> However, latencies > 1 sample are distorted only by an amount of < 1
> sample and they can hence be analyzed. So, for example, if two signals have
> a latency of 10 samples, the true latency lies between 9 and 11 samples
> (assumed that the latency is accurately estimated).
>
> Thanks and kind regards,
> Rikkert Hindriks
>
> ___
> HCP-Users mailing list
> HCP-Users@humanconnectome.org
> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>

___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

Re: [HCP-Users] Creating Custom dlabel File

[Timothy Coalson]

Creating a label file from ROIs is a bit more complicated than a single
command (label files automatically have a guarantee that the areas don't
overlap, but arbitrary ROIs can overlap).

>From what you currently have, one way to get to a new dlabel file is to
concatenate the ROIs you want to use in your desired order using
-cifti-merge, then use -cifti-reduce to generate both the INDEXMAX and MAX
reductions, use -cifti-math to zero out the INDEXMAX values wherever MAX is
zero (that is, everything that is unlabeled), and finally, use
-cifti-label-import (you will want to make a text file containing your
selected label names and desired colors).  One drawback of this is that the
label keys in the original MMP (say, 181 for L_V1) won't match the ones in
your dlabel file (they will go from 1 to the number of chosen areas, in the
order you picked).

Another way to approach it, especially if you have only a few areas you
want to keep or to remove, is to start over from the original, do
-cifti-label-export-table to save the original colors and names, use
-cifti-math on the dlabel file to zero out only the labels you don't want
(to zero out only R_V1 and L_V1, you can use the expression 'x * (x != 1 &&
x != 181)', see the text file from the first step to get the key values to
use, they are the first number in the line after the label name - as for
the expression, the part in parenthesis, because it comes from logical
operations, is always 0 or 1, so multiplying it into the original data (x)
is a masking operation - to do the opposite and keep only R_V1 and L_V1,
negate by inserting an exclamation point before the parenthesis: 'x * !(x
!= 1 && x != 181)').  The output of -cifti-math never has a label table,
and in this case will be a dscalar file, so you need to re-import it with
-cifti-label-import and the text file from the first step (and probably use
-drop-unused to take the removed areas out of the label table of the
result).

I am curious, why do you want a label file with fewer than the full 360
areas?

Tim


On Tue, Sep 5, 2017 at 5:08 PM, Timothy Hendrickson 
wrote:

> HCP users,
>
> I would like to create a custom dlabel file with particular ROIs from the
> Glasser 360 surface atlas.
>
> Here is what I have done so far:
>
> 1) First I separated individual ROIs into their dscalar files with
> wb_command -cifti-label-to-roi
> 2) I can easily visualize all dscalar files within wb_view
>
> Now I am stuck trying to take the individual ROI dscalar files and create
> one dlabel file which merges all ROI dscalar files together.
> I've tried several different commands including -cifti-create-dense,
> -cifti-create-dense-from-template, -cifti-create-label without luck.
>
> If anyone could assist it would be much appreciated.
>
> -Tim
>
> Timothy Hendrickson
> Department of Psychiatry
> University of Minnesota
> Bioinformatics and Computational Biology M.S. Candidate
> Office: 612-624-6441 <(612)%20624-6441>
> Mobile: 507-259-3434 <(507)%20259-3434> (texts okay)
>
> ___
> HCP-Users mailing list
> HCP-Users@humanconnectome.org
> http://lists.humanconnectome.org/mailman/listinfo/hcp-users
>

___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

Re: [HCP-Users] Maximum values for cognitive scores

[Elam, Jennifer]

Hi Ariana,
I did find these maximum values for the ASR raw scoring:

  *   ASR internalizing (maximum score for raw scores) = 78
  *   ASR externalizing (maximum score for raw scores) =70
  *   ASR anxiety and depression (maximum score for raw scores) = 36
  *   DSM depression (maximum score for raw scores) (from ASR) = 28
  *   DSM anxiety (maximum score for raw scores) (from ASR) = 14

I'm still working on trying to find the maximum values for the NIH Toolbox 
measures. You are right that scores available for download from ConnectomeDB 
are unadjusted scale scores and age-adjusted scale scores. If you are looking 
at the raw Toolbox data the scoring is not scaled and you may have to ask the 
Toolbox helpdesk for more information (h...@nihtoolbox.org). 

It may also be that since many of the Toolbox measures are computer-adaptive 
tests in which a subject's responses change what questions they are given for 
the rest of the test, there may not be a good answer for the maximum score 
possible. Looking at the Scoring Guide again, the Toolbox people would likely 
advise that you consider the scores scaled to their norming sample rather than 
an absolute max. In any case, I should be able to get you the highest scores 
that an HCP subject had for the tests for which you are interested.

Best,
Jenn

Jennifer Elam, Ph.D.
Scientific Outreach, Human Connectome Project
Washington University School of Medicine
Department of Neuroscience, Box 8108
660 South Euclid Avenue
St. Louis, MO 63110
314-362-9387
e...@wustl.edu
www.humanconnectome.org


From: Ariana Cahn 
Sent: Wednesday, August 30, 2017 4:22:59 PM
To: hcp-users@humanconnectome.org; Elam, Jennifer
Cc: Pascal Tetreault
Subject: Maximum values for cognitive scores

Hello,

I am trying to find the maximum scores for many of the cognitive/emotional 
tests administered to the HCP subjects. I have read the manual for the NIH 
toolbox and I cannot link their scoring system to the scores given by the HCP. 
From what I can tell, the HCP is providing the "Unadjusted Scale Scores", which 
is detailed on page 3 of the 2012 version of the NIH toolbox scoring and 
interpretation guide (attached in this email).
I was wondering what the maximum possible values are for the tests listed 
below. We have looked at the NIH toolbox manual and interpretation guide, the 
HCP data dictionary, and the ASEBA website (http://www.aseba.org/adults.html), 
but have not been able to find this information.

I need this information for the following scores:

  *   Picture Sequence
  *   List Sorting
  *   ASR internalizing (maximum score for raw scores)
  *   ASR externalizing (maximum score for raw scores)
  *   ASR anxiety and depression (maximum score for raw scores)
  *   perceived stress
  *   DSM depression (maximum score for raw scores)
  *   DSM anxiety (maximum score for raw scores)
  *   Positive Affect
  *   Self Efficiency
  *   Anger Affect
  *   Fear Affect
  *   Sadness Affect

Thanks,

--
Ariana Cahn
BSc (Hons), Neuroscience

ac...@ualberta.ca

___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

[HCP-Users] Creating Custom dlabel File

[Timothy Hendrickson]

HCP users,

I would like to create a custom dlabel file with particular ROIs from the
Glasser 360 surface atlas.

Here is what I have done so far:

1) First I separated individual ROIs into their dscalar files with
wb_command -cifti-label-to-roi
2) I can easily visualize all dscalar files within wb_view

Now I am stuck trying to take the individual ROI dscalar files and create
one dlabel file which merges all ROI dscalar files together.
I've tried several different commands including -cifti-create-dense,
-cifti-create-dense-from-template, -cifti-create-label without luck.

If anyone could assist it would be much appreciated.

-Tim

Timothy Hendrickson
Department of Psychiatry
University of Minnesota
Bioinformatics and Computational Biology M.S. Candidate
Office: 612-624-6441
Mobile: 507-259-3434 (texts okay)

___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

Re: [HCP-Users] clustering of subcortical structures - flipping and smoothing

[Miriam Klein-Flügge]

Dear all,

Thanks for these very helpful replies.

We can see that smoothing across boundaries will create strange effects 
at the edges (we had overlooked the 2mm parcel-constrained smoothing). 
Thanks for pointing that out. I am unsure from the below replies if 
there is a way to avoid having edges in the subcortical data (without 
going back to unprocessed data)?


The reason we would like to smooth the subcortical volume slightly more 
is because smoothing tends to help our clustering approaches. From the 
sound of it, we could easily apply clustering within a given subcortical 
parcel (with or without more smoothing) but it would not be valid once 
our area of interest to be clustered includes an edge and goes across 
parcels.


The swapping of hemispheres is a somewhat separate issue. We were hoping 
that both additional smoothing and potentially combining of 
(connectivity or similarity) data across hemispheres for clustering 
might help make the clustering more robust (if we are happy to ignore 
asymmetries for now), but it seems that the latter may create more 
problems than help.


Kind regards,

Miriam


On 01/09/2017 21:04, Timothy Coalson wrote:
I missed an important word: in the second paragraph, it should read 
"...the ideal way would be to also create an X flipped subcortical 
*label* volume file, then swap left and right labels in it..."


Tim


On Fri, Sep 1, 2017 at 3:02 PM, Timothy Coalson > wrote:


Using that transform that flips X around the origin should put
left on right and right on left, yes.  Note, however, that the
left and right parcel ROIs are not the same, so you will have to
figure out how to handle the mismatched edges.

I would not suggest -volume-parcel-resampling.  If you want it
back in cifti in the same subcortical ROIs, the ideal way would be
to also create an X flipped subcortical volume file, then swap
left and right labels in it (probably by
-volume-label-export-table, edit to exchange LEFT and RIGHT, then
-volume-label-import), then use -cifti-create-dense-* to make a
flipped cifti, then finally use -cifti-resample to deal with the
nonoverlapping ROI edges by resampling it back to the opposite
side original subcortical ROIs.  You could swap the left and right
surface data during this process also.

The -volume-parcel-smoothing command should not have an "across
parcels" option, that would be what "-volume-smoothing" is for. 
This option exists in cifti commands because cifti already
contains the parcel ROIs, and we decided the default would be to
treat them separately.

Tim


On Fri, Sep 1, 2017 at 2:11 PM, Glasser, Matthew
mailto:glass...@wustl.edu>> wrote:

I guess I don’t know why one would want to smooth across the
known boundaries, but the option is available in wb_command
-cifti-smoothing.  One could instead redo the volume mapping
to CIFTI with wb_command -volume-parcel-resampling using a
very small number for the kernel, and then apply the chosen
wb_command -cifti-smoothing afterwards.  There is a wb_command
-volume-parcel-smoothing that does not have the across parcel
option, though that would probably be trivial to add.

Matt.

From: "Harms, Michael" mailto:mha...@wustl.edu>>
Date: Friday, September 1, 2017 at 1:56 PM
To: Matt Glasser mailto:glass...@wustl.edu>>, Daria Jensen
mailto:dariajen...@gmail.com>>,
"hcp-users@humanconnectome.org
"
mailto:hcp-users@humanconnectome.org>>

Cc: Miriam Klein-Flugge mailto:mklein0...@googlemail.com>>
Subject: Re: [HCP-Users] clustering of subcortical structures
- flipping and smoothing

@Matt: The minimally preprocessed subcortical data in the
CIFTI was smoothed with a 2 mm FWHM “parcel-constrained”
kernel, right?  (i.e, the smoothing does not cross boundaries
of subcortical structures).  In that case, if they were to
then further smooth the subcortical data, but smooth across
subcortical boundaries (as they proposed), wouldn’t that
potentially lead to some odd effects at the boundaries?  [I
don’t know if you caught that part of their proposal…]

cheers,

-MH

-- 


Michael Harms, Ph.D.

---

Conte Center for the Neuroscience of Mental Disorders

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


*From: *mailto:hcp-users-boun...@humanconnectome.org>> on behalf of
"Glasser, Matthew" mailto:glass...@wustl.edu>>
*Date: *Friday, September 1, 2017 at 1:50 PM
 

Re: [HCP-Users] Slice-timing correction and latency structure of resting-state fMRI data

[Glasser, Matthew]

The TR=0.72s, but I think that is correct.

Peace,

Matt.

From: 
mailto:hcp-users-boun...@humanconnectome.org>>
 on behalf of "HINDRIKS, RIKKERT" 
mailto:rikkert.hindr...@upf.edu>>
Date: Tuesday, September 5, 2017 at 8:48 AM
To: "hcp-users@humanconnectome.org" 
mailto:hcp-users@humanconnectome.org>>
Subject: [HCP-Users] Slice-timing correction and latency structure of 
resting-state fMRI data


Dear all,

I am analyzing the latency structure of some of the HCP resting-state fMRI data 
and I want to make sure that this makes sense, given that no slice-timing 
correction has been applied to the data. I would appreciate it if someone could 
confirm that the following reasoning is correct:

Since the entire brain is scanned in 0.78 seconds (one sample), it does no make 
sense to analyze signals latencies < 1 sample, because such small latencies 
will be distorted. In particular, it does not make sense to  interpolate the 
cross-covariance functions as done in one of Mitra's papers).

However, latencies > 1 sample are distorted only by an amount of < 1 sample and 
they can hence be analyzed. So, for example, if two signals have a latency of 
10 samples, the true latency lies between 9 and 11 samples (assumed that the 
latency is accurately estimated).

Thanks and kind regards,
Rikkert Hindriks


___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users

[HCP-Users] Slice-timing correction and latency structure of resting-state fMRI data

[HINDRIKS, RIKKERT]

Dear all,

I am analyzing the latency structure of some of the HCP resting-state fMRI
data and I want to make sure that this makes sense, given that no
slice-timing correction has been applied to the data. I would appreciate it
if someone could confirm that the following reasoning is correct:

Since the entire brain is scanned in 0.78 seconds (one sample), it does no
make sense to analyze signals latencies < 1 sample, because such small
latencies will be distorted. In particular, it does not make sense to
 interpolate the cross-covariance functions as done in one of Mitra's
papers).

However, latencies > 1 sample are distorted only by an amount of < 1 sample
and they can hence be analyzed. So, for example, if two signals have a
latency of 10 samples, the true latency lies between 9 and 11 samples
(assumed that the latency is accurately estimated).

Thanks and kind regards,
Rikkert Hindriks

___
HCP-Users mailing list
HCP-Users@humanconnectome.org
http://lists.humanconnectome.org/mailman/listinfo/hcp-users