Resection identification in Brainstorm #848

chinmaychinara91 · 2025-10-07T00:14:11Z

This PR integrates the resection-identification from https://github.com/ajoshiusc/auto_resection_mask/tree/brainstorm-plugin as a Brainstorm plugin. Given a pre and post op MRI, this generates two resection masks:

Postop MRI resection mask
Postop MRI resection mask warped in the preop MRI space. ~~(Discussed this with @ajoshiusc and this is the most relevant one that is used in the community, so I am just importing that).~~
EDIT: As discussed on Slack, we import both the masks (as atlases) and the post2preop coregistered MRI from the plugin outputs.
EDIT2: It is also relevant that we generate the resection volume from the masks.

Test data: Example data from the repo above.

The process can be found as seen in the image below. The process requires choosing the preop and postop MRI NIfTI files using the file browser.

If pre and postop MRI already present in a subject, one can ~~select both of them,~~ right-click on the postop and start the resection identification as seen below.

Tutorial (WIP): https://neuroimage.usc.edu/brainstorm/Tutorials/ResectionIdentification

IFF exactly two MRIs are selected. 1st is the pre-op MRI 2nd is the post-op MRI

rcassani

It looks fine, there are some details to address:

Why is it called labeling? There is no labeling, but masking, or resection identification.
Is there a plotting in calling resection_labeling?
I got the messages that some plotting was happening but the figures did not show up. If there is plotting, can we move it to the process code so it plots Brainstorm figures?

EDIT (Add the message showed in the Matlab command window about plotting)

epoch_loss: �[94m -0.0753 �[0m  for epoch: �[92m 2999/3000 �[0m 
nilearn/plotting/find_cuts.py:154: UserWarning: Could not determine cut coords: All voxels were masked by the thresholding. Returning the center of mass instead.
  warnings.warn(
Plotted preop MRI with resection mask for preop.nii.gz
Plotted postop MRI with resection mask for postop.nii.gz

See the other individual comments

toolbox/process/functions/process_resection_identification.m

toolbox/process/functions/process_resection_labeling.m

toolbox/tree/tree_callbacks.m

rcassani · 2025-10-07T15:50:46Z

toolbox/process/functions/process_resection_identification.m

+        bst_report('Error', sProcess, [], ['Subject "' SubjectName '" is using the default anatomy (read-only).']);
+        return
+    end
+    % Proceed importing the MRI files only if no anatomy defined for subject


Do we want this limitation?

A way to give the user the option to do the same thing with GUI and with Processes would be to:

Modify process_import_mri to add an option isAutoAdjust (I'll handle this in master)

Then the pipeline would look:

Process IMPORT MriFile1 as MRI1 Process IMPORT MriFile1 as MRI2 Process RESECT(MRI1, MR2)

Your suggestion sounds much better. Definitely, now it's better to not have this limitation.

The process process_import_mri() handles now CT and PET volumes. Check b4b2da3

So, the user can give either the FileName or the Comment of the PostOp MRI.
And the resection identification will happen against the Default MRI for that Subject

So, the user can give either the FileName or the Comment of the PostOp MRI.
And the resection identification will happen against the Default MRI for that Subject

From the process_resection_identification, instead of asking for both the pre-op and post-op (as it is now) we just ask for post-op MRI (image below) as it is expected that the pre-MRI is already imported/set as the default MRI ?

If pre-MRI is not present then we give an appropriate message to user that the pre-MRI needs to be imported first. Does this sound good ?

What if PostOp MRI is already imported?
(This is the case as calling from the db tree)

~~In that case, like we have the dropdown list for the Subject in the image above, we can have a dropdown list for choosing the post-op MRI from all the non-default MRI volumes~~
EDIT: How about just defining a text field for post-op MRI name in the process tab, where user can just type in the name of the postop file from the db tree ? Check e2f3e4d.

EDIT: How about just defining a text field for post-op MRI name in the process tab, where user can just type in the name of the postop file from the db tree ?

This was suggested in #848 (comment)

rcassani · 2025-10-07T20:22:44Z

toolbox/core/bst_plugin.m

+    switch(OsType)
+        case 'linux64'
+            PlugDesc(end).URLzip   = 'https://neuroimage.usc.edu/bst/getupdate.php?d=bst_resection_labeling_linux.zip';
+            PlugDesc(end).TestFile = 'resection_labeling';


Need to add execute permission to resection_labeling (linux64 and mac64*)

Maybe we should have it directly embedded in the binary ? Let me check this.
EDIT: Handled in the binary. Associated PR: ajoshiusc/auto_resection_mask#4

toolbox/process/functions/process_resection_identification.m

chinmaychinara91 · 2025-10-08T04:59:54Z

Why is it called labeling? There is no labeling, but masking, or resection identification.

I just followed the naming as per the tutorial https://neuroimage.usc.edu/brainstorm/Tutorials/SegBrainSuite#Resection_labeling but agree resection_identification sounds good or maybe resection_mask_id. Let me know which of these sounds better and I will proceed with generating binary accordingly.
EDIT: Going ahead with resection_identification as it sounds more appropriate

Is there a plotting in calling resection_labeling?
I got the messages that some plotting was happening but the figures did not show up. If there is plotting, can we move it to the process code so it plots Brainstorm figures?

There is no plotting actually that needs to be displayed rather the plotting happens to generate two image files (below) for sanity check on the resection mask generation. Don't think we need to have them as Brainstorm figures as once the resection mask is imported we can overlay it on the preop MRI and check it.

This custom directory is the path where the binary packages will be extracted. If not specified, it uses the system's temporary directory as the default.

…olume

- If it is an empty subject, import the pre-op, do MNI normalization, and then import the post-op. - If already anatomy present in the subject, then just import them as two additional MRI files to the subject. Since, `isAutoadjust` is set to 1, so they will reslice/resample based on the default MRI)

- The preop and postop MRIs need to be imported first to a subject before running the process. - The pre-op MRI is always the default MRI so no need to explitly ask for it in the process tab. - The post-op MRI could be any of the non-default MRI. To keep it simple, in the process tab user can type in the name of it as in the db tree.

rcassani

Looks good

Few extra comments:

The permissions to be executable does not seem to be properly set, at least in Linux. Test: download and unzip the .zip file
There is still the legend about plotting, that one is confusing as the generated images are not shown, not even saved. Can you remove the generation of those images from the binaries?

nilearn/plotting/find_cuts.py:154: UserWarning: Could not determine cut coords: All voxels were masked by the thresholding. Returning the center of mass instead.
  warnings.warn(
Plotted preop MRI with resection mask for preop.nii
Plotted postop MRI with resection mask for postop.nii
RESEC_ID > Computation completed in: 1256.6037s

About the menu entry.
Does it fit better under MRI segmentation?

toolbox/core/bst_plugin.m

toolbox/process/functions/process_resection_identification.m

rcassani · 2025-10-17T18:57:53Z

toolbox/process/functions/process_resection_identification.m

+        bst_report('Error', sProcess, [], ['Subject "' SubjectName '" is using the default anatomy (read-only).']);
+        return
+    end
+    % Proceed importing the MRI files only if no anatomy defined for subject


EDIT: How about just defining a text field for post-op MRI name in the process tab, where user can just type in the name of the postop file from the db tree ?

This was suggested in #848 (comment)

chinmaychinara91 · 2025-10-21T08:51:29Z

The permissions to be executable does not seem to be properly set, at least in Linux. Test: download and unzip the .zip file

The permissions are set now. I set them directly in the server.

There is still the legend about plotting, that one is confusing as the generated images are not shown, not even saved. Can you remove the generation of those images from the binaries?

Updated the binaries to remove the generation of the images.

About the menu entry.
Does it fit better under MRI segmentation?

Definitely a better fit. Check 98d7f19

1. Remove repeated check for MRIs to be from the same subject and not default anatomy from `Compute` because the checks are already done when calling from GUI. 2. Make sure the fiducials in pre-op MRI are set as the surface (from the resection masks) needs the SCS coordinates. 3. Add `ComputeInteractive` function.

chinmaychinara91 added 3 commits September 30, 2025 12:18

Add resection-labeling plugin

d275369

Initial commit: process_resection_labeling

015546d

Right click and call process from anatomy

f8a307a

IFF exactly two MRIs are selected. 1st is the pre-op MRI 2nd is the post-op MRI

chinmaychinara91 marked this pull request as draft October 7, 2025 00:14

chinmaychinara91 added 3 commits October 7, 2025 00:58

Clean up

f1fddc2

Clean up

421c1da

Rename to maintain uniformity

b5899eb

chinmaychinara91 marked this pull request as ready for review October 7, 2025 08:37

rcassani requested changes Oct 7, 2025

View reviewed changes

Typo: Should be ICON_FEM

a296c90

chinmaychinara91 requested a review from rcassani October 8, 2025 06:35

chinmaychinara91 marked this pull request as draft October 9, 2025 21:14

Plugin: Rename resection-labeling to resection-identification

ebe893c

chinmaychinara91 changed the title ~~Resection labeling in Brainstorm~~ Resection identification in Brainstorm Oct 13, 2025

chinmaychinara91 added 10 commits October 14, 2025 00:34

The plugin now allows uncompressed NIfTi inputs

b510888

Confirmation not required for deleting

8cf2ab0

Pass the custom directory argument

9a58277

This custom directory is the path where the binary packages will be extracted. If not specified, it uses the system's temporary directory as the default.

Typo

94919bd

Resection done against the Default MRI

dfe3e5c

Save outputs: pre- and post-op masks and post2pre non-linear coreg. v…

fde996c

…olume

Update usage and iSubject not required

968b163

Same subject and no default anatomy check

53e160e

chinmaychinara91 marked this pull request as ready for review October 17, 2025 10:10

rcassani reviewed Oct 17, 2025

View reviewed changes

chinmaychinara91 added 3 commits October 20, 2025 14:25

Update standalone binaries for the plugin

bb8aec5

Not required

1157dcf

Search by Comment or FileName

8020dad

chinmaychinara91 added 5 commits October 21, 2025 00:52

isOk flag to track status of process

b48563f

Computation time: Round to nearest integer

5d7e80c

Use PlugDesc.TestFile

d4bc2ba

sSubject.iAnatomy to check if default MRI is present

200f9cd

Menu entry moved under MRI segmentation

98d7f19

chinmaychinara91 added 3 commits October 21, 2025 03:17

Bugfix: MriFile refers to the post-op MRI

f42168f

Process: Handle pre-op MRI name in pos-op field

8489e74

chinmaychinara91 requested a review from rcassani October 21, 2025 10:46

rcassani added 3 commits October 21, 2025 10:45

Bugfix

32bde3b

Improve error messages

b298d78

Keep it simple

711795d

Resection identification in Brainstorm #848

Are you sure you want to change the base?

Resection identification in Brainstorm #848

Conversation

chinmaychinara91 commented Oct 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rcassani left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rcassani Oct 7, 2025

Choose a reason for hiding this comment

Uh oh!

chinmaychinara91 Oct 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rcassani Oct 16, 2025

Choose a reason for hiding this comment

Uh oh!

rcassani Oct 16, 2025

Choose a reason for hiding this comment

Uh oh!

chinmaychinara91 Oct 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rcassani Oct 16, 2025

Choose a reason for hiding this comment

Uh oh!

chinmaychinara91 Oct 16, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rcassani Oct 17, 2025

Choose a reason for hiding this comment

Uh oh!

rcassani Oct 7, 2025

Choose a reason for hiding this comment

Uh oh!

chinmaychinara91 Oct 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

chinmaychinara91 commented Oct 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rcassani left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

rcassani Oct 17, 2025

Choose a reason for hiding this comment

Uh oh!

chinmaychinara91 commented Oct 21, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

chinmaychinara91 commented Oct 7, 2025 •

edited

Loading

rcassani left a comment •

edited

Loading

chinmaychinara91 Oct 8, 2025 •

edited

Loading

chinmaychinara91 Oct 16, 2025 •

edited

Loading

chinmaychinara91 Oct 16, 2025 •

edited

Loading

chinmaychinara91 Oct 8, 2025 •

edited

Loading

chinmaychinara91 commented Oct 8, 2025 •

edited

Loading