Lesion in MNI to CIFTI

[amirhussein.a]

Dear HCP users/experts,

We are interested in calculating the functional connectivities of a lesion mask, which has been drawn in MNI space. The lesion involves both cortical and subcortical areas. These are the steps I took to find the associated CIFTI-indices of the lesion to use and calculate the functional connectivity of the lesioned areas with other cifti greyordinate time-series data. This is the pipeline:

1. Make a "lesion.txt" labeling value 1 as "Lesion" and color code it into the red.
2. -volume-label-import  lesion_MNI.nii   lesion.txt    lesion_wb.nii  #import into WorkBench format
3. -volume-label-to-surface-mapping   lesion_wb.nii   groupavg_midthickness.[L/R].gii  lesion_[L/R].label.gii   -ribbon-constrained  groupavg_white.gii  groupavg_pial.gii #Map to surface
4. -cifti-create-label   -volume  lesion_wb.nii  Atlas_ROIs.2.nii.gz  -left-label   Lesion_L.label.gii  -right-label   lesion_R.label.gii   lesion.dlabel.nii  #Map to volume and merge with surface labels

The final lesion.dlabel.nii can be opened by workbench and clearly shows lesions on the cortical surface and subcortical volumes (visually exactly what I expected). But, it has 96k indices when I open with ciftiopen in MATLAB, possibly due to the inclusion of medial wall in volume label to surface mapping. Thus, it is hard if not impossible to map it with 91k greyordinate time-series data. I chose another strategy by changing the target surface for -volume-label-to-surface-mapping:

1. Download https://github.com/Washington-University/HCPpipelines/blob/master/global/templates/91282_Greyordinates/91282_Greyordinates.dscalar.nii
2. -cifti-separate 91282_Greyordinates.dscalar.nii COLUMN -metric    CORTEX_[RIGHT/LEFT]    [right/left]cortex.shape.gii
3. Open the resulting cortex.shape.gii files using gifti tools in MATLAB (32k with 1 for not medial wall and 0 values for medial wall), open the groupavg_midthickness.surf.gii, dot product them (cortex.cdata .* midthickness.vertices), remove the rows with 0 values, change the "faces" of midthickness manually (I couldn't find any description on what faces correspond to), saving rows 1:59105 for right and 1:59065 for left hemisphere .vertices, save them into new midthickness files.
4. Repeat the steps 3 and 4 in the previous part

Here we will have ~91k indices, that can be matched row-wise with dtseries files. However, when I try to visualize them here is the error I get from workbench:

"Surface file contains 29696 vertices but the CortexLeft contains 32492 vertices."

I believe that I have done everything correct (am I?) and I can proceed to calculate functional connectivity of the lesion mask with other greyordinates in dtseries file. But, I'd like to visualize the label file I have generated, to visually check whether it has been mapped correctly or not.

Any comments regarding the pipeline or how to correct the error I encountered is appreciated.

Best,

AmirHussein Abdolalizadeh

[Glasser, Matt]

A lesion suggests that this is from an individual subject.  I would map to that individual’s surface and subcortical CIFTI space rather than attempting a group mapping.

 

I’m also not clear on the purpose of estimating functional connectivity from a “lesion” so that part should be clarified also, off list if necessary as to why you are doing this. 

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/e686a318-4b39-4be3-8b5c-64e6ce7336aen%40humanconnectome.org.

 

---

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.

[amirhussein.a]

Hi Matt. Thanks for your reply.

We intended to do a lesion network mapping. We have a group of subjects who have had brain traumatic injuries with various degrees and locations but share a common behavioral consequence. By seeding the lesioned area that has been transformed into an atlas space (e.g., MNI) in a large healthy unaffected sample, we try to identify which common connections/networks may have been affected by the lesion that can explain the change in behavior.

Now, we have lesion masks in the MNI space for our subjects drawn carefully by an expert neuroradiologist based on their CT scans (lesion1, lesion2, etc.). We decided to find the CIFTI-indices associated with each individual mask, and run a functional connectivity analysis seeding the mask in all HCP participants (a.k.a., our healthy sample) on their rsfMRI preprocessed greyordinates, average/smooth, and find the affected common connections.

[Coalson, Timothy Scott (S&T-Student)]

No, changing the number of gifti vertices not the intended way to do this, and the volume mapping isn't the cause of either problem.  I would recommend -cifti-create-dense-from-template instead of -cifti-create-label, but -cifti-create-label does have options to take those cortical ROIs as inputs.

Assuming the lesions interfere with surface generation or identifying the affected location on healthy surfaces that way, you should consider mapping your ROIs to at least a handful of randomly-chosen healthy subjects to get a probability distribution of the ROI location, as the group average surfaces do not follow the cortical contours very well, see figure 9 and supplementary figures S1 and S9 in our paper https://www.pnas.org/doi/full/10.1073/pnas.1801582115 .

Tim

---

From: amirhussein.a <amirhu...@gmail.com>
Sent: Saturday, May 7, 2022 3:23 PM

To: HCP-Users <hcp-...@humanconnectome.org>
Subject: [hcp-users] Lesion in MNI to CIFTI

 

WARNING: This message has originated from an External Source. This may be a phishing expedition that can result in unauthorized access to our IT System. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.

--

[amirhussein.a]

Hi Tim, thanks for your reply and heads up.

To check whether I have understood correctly or not:

• Apply -volume-label-to-surface-mapping using each HCP subjects' left/right midthickness files in /MNINonLinear/fsaverage_LR32k: R or L_midthickness.32k_fs_LR.surf.gii (NOT the _MSMALL ones?)
  
• -cifti-create-dense-from-template using the resulting label.gii files of each HCP subject and the template will be the 91282_greyordinates
• Run rest fMRI connectivity analysis for each individual rsfMRI scan seeding their individualized label files.

Right?

[Coalson, Timothy Scott (S&T-Student)]

1. If you are interested in what functional areas "should have been" where the lesion now is, then the MSMAll surfaces would be appropriate, as they would capture the additional uncertainty from where functional areas typically land with respect to folding patterns.  You don't need to use all HCP subjects, but I don't have data to back up how many is "enough".  Probably more than 10.

2. Yes.

3. I think we have (healthy) group average dense connectivity available for HCP YA already (depending on how our filesystem recovery is going...), so you shouldn't need to run your own.  We used MSMAll for our processing, which immediately suggests that you should use the MSMAll surfaces for step 1.

Tim

---

From: amirhussein.a <amirhu...@gmail.com>
Sent: Wednesday, May 18, 2022 5:04 AM
To: HCP-Users <hcp-...@humanconnectome.org>
Cc: Coalson, Timothy Scott (S&T-Student) <tsc...@mst.edu>
Subject: Re: [hcp-users] Lesion in MNI to CIFTI

 

[amirhussein.a]

Great, thanks!

Bests,

Amir