Focality optimisation step 2/2 fails: 'FlexSearchTab' has no attribute 'volume_atlas_display_map'

[LianneKeighery]

TI-Toolbox version: v2.2.4
OS: Windows (Dell)
Interface: Desktop GUI
Date encountered: April 2026

Hi, thanks so much for creating this brilliant tool! I am running group-level simulations using the Ernie template head model but encountering the below error message in the output in the Optimizer step and wondered if you may be able to help please?

Error:
Error calculating adaptive thresholds: 'FlexSearchTab' object has no attribute 'volume_atlas_display_map'

As a key goal of mine is to maximise TI field intensity in the right amygdala (ROI, subcortical label 54) while minimising field spread in the right hippocampus (non-ROI, subcortical label 53), I've set my optimization goal to Focality (maximise field in ROI, minimise elsewhere). See below a summary of the settings I selected:

Optimiser settings:

• Optimization Method: Flex-Search; Weise et al. 2025
• Goal: Focality (maximise field in ROI, minimise elsewhere)
• ROI: subcortical label 54 (right amygdala)
• Non-ROI Definition Method: Specific Region
• Non-ROI Volume Atlas: currently left as blank / unselectable. Is this a glitch, or it's supposed to remain blank? When I click on it, no options appear (see screenshot )
• Non-ROI Volume Label: 53 (right hippocampus)
• Use Adaptive Thresholds:

enabled - Non-ROI adaptive threshold: 20% - ROI adaptive threshold: 80%

[idossha]

Hi @LianneKeighery,

Happy you like the TI-Toolbox! We hope it can help accelerate your study.

Regarding this issue - I believe this was fixed with the latest release (2.3.0).

If you need to use 2.2.4 specifically, I can work on patching a fix for you, otherwise would you consider upgrading to the latest?

As a heads up - there were some breaking of backward compatibility between those versions, but if you are not deep in your optimization process I would suggest upgrading. There are some useful features in 2.3.0.

Let me know what you end up doing. Always happy to assist or brainstorm together.

Best,
Ido

[LianneKeighery]

Hi Ido, Thanks so much for the quick response! I've gone ahead and updated to v2.3.0. I just wanted to check, given the breaking changes between versions, does this mean I need to re-run my preprocessing, optimizer and simulation steps from scratch, or is there any way to carry over my existing work from v2.2.4? Thanks again! Best wishes, Lianne

…

___________________________________________________________ Lianne Keighery | PhD candidate Dept. of Clinical and Movement Neurosciences Address: UCL Institute of Neurology | 33 Queen Square | University College London (UCL) Email: ***@***.*** Website: UCL Institute of Neurology<https://www.ucl.ac.uk/ion/ucl-queen-square-institute-neurology> [cid:31f67686-c42f-42a4-9416-e65a0a6d51d2] [cid:34b07ff9-b0f0-4ff1-8067-b021797725a7] [cid:989857d6-0eb4-44ad-9ecc-9000eff51fca] I am a Mental Health First Aider. You can talk to me about a mental health issue and I can provide initial support and signpost you to appropriate help. Don’t suffer in silence. Supporting Being Well at UCL<https://www.ucl.ac.uk/human-resources/health-wellbeing/being-well-ucl>

________________________________ From: multipass ***@***.***> Sent: 16 April 2026 5:07 PM To: idossha/TI-Toolbox ***@***.***> Cc: Lianne Keighery ***@***.***>; Mention ***@***.***> Subject: Re: [idossha/TI-Toolbox] Focality optimisation step 2/2 fails: 'FlexSearchTab' has no attribute 'volume_atlas_display_map' (Discussion #89) ⚠ Caution: External sender Hi @LianneKeighery<https://github.com/LianneKeighery>, Happy you like the TI-Toolbox! We hope it can help accelerate your study. Regarding this issue - I believe this was fixed with the latest release (2.3.0). If you need to use 2.2.4 specifically, I can work on patching a fix for you, otherwise would you consider upgrading to the latest? As a heads up - there were some breaking of backward compatibility between those versions, but if you are not deep in your optimization process I would suggest upgrading. There are some useful features in 2.3.0. Let me know what you end up doing. Always happy to assist or brainstorm together. Best, Ido — Reply to this email directly, view it on GitHub<#89 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BVUZVTCWHW377C3ZIY3SNFT4WEAMDAVCNFSM6AAAAACX3LERACVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMNJYHE3DOMA>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[idossha]

Preprocessing should be all good.

finished optimizer - simulations pairs should also be all good.

The only difference is in the naming of the directories of the flex-search output. Which means, if you want to rerun simulations based off of previous flex-search outputs, it might not work.

The bottom line, you should be able to continue as usual, however, some functions might be yelling at out. Nothing quantitative changed from v2.2.4 -> v2.3.0, it was mostly file/dir organization changes.

Just be aware of that.

[LianneKeighery]

Hi Ido, Great, thank you so much for your help! I have v 2.3.0 up and running and all is going well so far. I had another question, please. When visualising and analysing my output, is there a function to view 2 simulations concurrently at the Analyzer stage? As an example, if I set my target ROI as the right amygdala using the focality optimization goal and set a specific region (i.e., right hippocampus) as my non-ROI site, and ran a second optimization in the right amygdala using the focality optimization goal , but this time specified my Non-ROI Definition Method as "Everything Else (Default)", is it possible to pull up both in the same plots for visualization, so I can do a quick comparison of electric field distribution and values in each? Many thanks for your help. Best wishes, Lianne

…

___________________________________________________________ Lianne Keighery | PhD candidate Dept. of Clinical and Movement Neurosciences Address: UCL Institute of Neurology | 33 Queen Square | University College London (UCL) Email: ***@***.*** Website: UCL Institute of Neurology<https://www.ucl.ac.uk/ion/ucl-queen-square-institute-neurology> [cid:ef9004fe-22f4-44f4-986e-996951f76edf] [cid:504da95b-c11e-43ae-836b-e0569651fe59] [cid:addc1bd9-b642-41f0-8f13-56cbe20a9e06] I am a Mental Health First Aider. You can talk to me about a mental health issue and I can provide initial support and signpost you to appropriate help. Don’t suffer in silence. Supporting Being Well at UCL<https://www.ucl.ac.uk/human-resources/health-wellbeing/being-well-ucl>

________________________________ From: multipass ***@***.***> Sent: 20 April 2026 11:57 PM To: idossha/TI-Toolbox ***@***.***> Cc: Lianne Keighery ***@***.***>; Mention ***@***.***> Subject: Re: [idossha/TI-Toolbox] Focality optimisation step 2/2 fails: 'FlexSearchTab' has no attribute 'volume_atlas_display_map' (Discussion #89) ⚠ Caution: External sender Preprocessing should be all good. finished optimizer - simulations pairs should also be all good. The only difference is in the naming of the directories of the flex-search output. Which means, if you want to rerun simulations based off of previous flex-search outputs, it might not work. The bottom line, you should be able to continue as usual, however, some functions might be yelling at out. Nothing quantitative changed from v2.2.4 -> v2.3.0, it was mostly file/dir organization changes. Just be aware of that. — Reply to this email directly, view it on GitHub<#89?email_source=notifications&email_token=BVUZVTFZRXSK6J7TB7Z3BG34W2TMRA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNRWGQZTKNBTUZZGKYLTN5XKO3LFNZ2GS33OUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#discussioncomment-16643543>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/BVUZVTFBTAT46OIR3O2CNOT4W2TMRAVCNFSM6AAAAACX3LERACVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTMNRUGM2TIMY>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[idossha]

Absolutely. We support this need via the nifti visualizer tab.

Go to group mode and load the two different simulations. It should automatically overlay them.

I am working on updating our Video tutorials page, and I will make sure to cover comparisons like this one. Hope that helps for now.

p.s. If you find the TI-Toolbox helpful for your study - feel free to send us feedback and feature requests or report bugs so we can continue and support your research needs. Starring this repository will also help us continue to grow!

Best regards,
Ido

[LianneKeighery]

Hi Ido,

Thanks so much for your response, your guidance has been so helpful for accelerating my workflow after hitting some obstacles.

I wanted to flag a follow-up issue I encountered after reading through another discussion regarding Flex-Search electrode outputs and Gmsh visualization on GitHub (#73). Based on your answer there, I expected to find a highlighted electrode visualisation PNG at the following path after running the mapping approach:

/{project_name}/derivatives/SimNIBS/sub-{id}/Simulations/{simulation_name}/TI/montage_imgs/{simulation_name}_highlighted_visualization.png

However, this PNG file does not exist in my project directory and my montage_imgs folder is empty. I can confirm that the Flex Search Optimizer and Simulator steps completed successfully using the mapping approach so I am not sure whether this file is failing to generate silently, or whether there is a configuration step I may have missed. Any guidance on what might cause this output to be absent would be very helpful.

I also had a question regarding target ROI definition. My study ideally requires stimulation of both the left and right amygdala concurrently. I am interested in whether it would be possible to define a bespoke combined bilateral ROI (i.e. both amygdalae as a single target) rather than running them as two separate single-target optimisations. Do you have any recommendations for how best to approach this within the current toolbox framework, for example, whether a merged ROI mask could be passed to the Flex Search optimizer, or whether there is another approach you would suggest for bilateral subcortical targeting?

I would be very happy to discuss my optimisation goal needs in more depth if it helps to clarify things. If you are open to it, would it be possible to schedule a short call to discuss this?

Thank you again for your support! It's much appreciated!

Best wishes,
Lianne

[idossha]

Hi Lianne,

Happy to get on a Zoom call and discuss if you want to brainstorm. Just give me a few options, I am generally flexible. Send me an email if relevant.

• Regarding bi-lateral flex-search. This is impossible with uni-polar montages. Unless you do care about focality. You could try to create a sphere that is large enough to encompass both ROIs.
• Regarding image: not sure why this is not working for you. This should be automatically produced after running a simulation with either a “montage” or “mapped” configurations.

To be clear - the img is produced after running the simulation from the simulation tab. Not from within the flex-search tab.

Ido

[LianneKeighery]

Hi Ido,

Thank you so much for your willingness to discuss further. I will send a separate email to suggest some options.

Regarding bi-lateral flex-search. This is impossible with uni-polar montages. Unless you do care about focality. You could try to create a sphere that is large enough to encompass both ROIs.

• Understood that this isn't possible with unipolar montages, and I did worry that this was an overly ambitious goal. As focality is particularly important for my study (I want to avoid hippocampal area as much as possible), I was considering running a multipolar approach so that I can target two amygdalae (right and left) using 2 electrode pairs to produce 2 x envelopes Δf₁ and Δf₂ in an (i) interleaved (R,L,R,L), (ii) sequential (R,R,L,L) or (iii) simultaneous (RL, RL) fashion. However given the proximity and small volume of two amygdalae would I be right in saying that Δf₁ and Δf₂ must differ from each other to avoid inter-system interference producing spurious envelopes elsewhere? For example, if the two difference frequencies are kept distinct (e.g., left: Δf = 10 Hz, right: Δf = 20 Hz), might this account for this issue?

Regarding image: not sure why this is not working for you. This should be automatically produced after running a simulation with either a “montage” or “mapped” configurations.

• I've deleted the folder directory for my simulation and re-run it and am still encountering the same issue. Perhaps we could revisit this issue on my call if I talk you through my workflow.

Another separate issue I've spotted is that when I select 2 distinct optimization goals in the Simulator tab -> Montage / Flex Search Selection - with slight variations in the Non-ROI parameters (one specifies right Amyg as target ROI; right HPC as the non-ROI region, whilst the other specifies right Amyg as target ROI; everything else as the non-ROI region), the resulting output folder is identical (see screenshot) and seems to get overridden. Folder name here is flex_focality_max_TI_mapped however it seems to only contain one simulation within each subfolder - not two - though I have ran both.

Is it possible to distinguish this more clearly for the purpose of clearer navigation in my project directory?

Further details of what I ran at Optimization stage below for each of these:
Optimization Goal:

1. focality (maximise field in target ROI while minimizing field elsewhere) with ROI target right amygdala, **non-ROI definition method = specific region (right hippocampus)
   VS
2. focality (maximise field in target ROI while minimizing field elsewhere) with ROI target right amygdala, non-ROI definition method = everything else (default)

Post-processing method
max_TI (maximum TI field)

Anisotropy type
Isotropic (scalar)

ROI Definition

• Subcortical
• Volume atlas: labeling.nii.gz
• Tissue type: gray matter

Focality options
Adaptive (single run)
Adaptive percentages (non ROI = 20%, non ROI = 80%)

Thank you again for being so helpful and proactive in your responses!

Best wishes,
Lianne

[idossha]

Hi @LianneKeighery — quick follow-up. The specific focality step 2/2 GUI lifecycle issue has been further hardened for v2.3.1: the adaptive focality step no longer marks itself finished before the backend process actually completes, and GUI tabs now use subprocess success/failure status more consistently.

For your separate comparison question: a dedicated side-by-side Analyzer comparison view is not in v2.3.1 yet, but the Analyzer and NIfTI Viewer refresh behavior has been improved so newly generated simulations/analyses are easier to select immediately after running. I’ll keep the concurrent-simulation comparison as a feature request.

These fixes should be included in v2.3.1 in the coming days.