This question may be directed to Dr Thomas Vincent,
We’re using NIRStorm’s process “Compute optimal montage” to preplan the optode placements for our cognitive experiments. First of all, thank you for making the fluence files available on your website (thomasvincent.xyz). It takes a very long time for us to download all the files (over 7000 of them for our prefrontal regions of interest). After 24 hours, only about 1500 files had been downloaded.
Would it be possible to download all the fluence files at once, e.g. in a zip file?
no, unfortunately it is not possible to generate such zip file. Could you explain how you created the ROI for your prefontal region and how you are launching the optimal montage ? it sounds weird to me that you need more than one or two thousands fluences .
Here are the ROIs. The main cognitive tasks in our experiments are the 2-Back and stroop tasks, therefore we'd like to cover the wider dorsolateral PFC. Possibly not as superior as the figure shows below (top of pink superior gyrus). But I simply used the Destrieux atlas with the following regions. I see that there are 4006 vertices, so that means 4006 fluence files?
But because I have all files now locally, this process completed within about 5-10 minutes.
Here are the results, and that is new for me too this morning. The optimal position of optodes is limited to one hemisphere, and is also limited to the more ventral parts of the ROIs. Not the result I was expecting. I must be doing something wrong, e.g. do I need to combine the multiple ROIs into one scout?
indeed, there are several issues here. mainly due to the lack of documentation of the process (actually, if you would like to help for writing some tutorials, that would be very helpful :))
Regarding the ROI selection:
-> Currently, the optimal montage in NIRSTORM can only use one scout at a time. Here, when you selected multiple ROI, NIRSTORM only considered the first one. Being able to work with multiple ROIs is on my to-do list but I don't know when it will be done.
There are two solutions currently: merging the ROI and using it for the optimum montage. Or doing one optimal montage per ROI and then merging the montages (you have a process called merge montage).
There are issues with the two methods:
If you merge the ROI, you will end with an extended target region; or the optimal montage in NIRSTORM has mainly been developed to work for the focal region (mainly for epilepsy and doesn't perform very well on the extended region. The reason is that we are maximizing the sensitivity of the montage to the ROI without any constraint on the coverage of the ROI. (this is also something we are currently investigating)
If you apply OM on the regions individually, merging might be difficult as you might have overlap optodes.
-> Regarding the Head Scout selection: This is similar to the issue described at the end of the thread here: (https://github.com/Nirstorm/nirstorm/issues/183); In head scout, you should select a region describing where you can put optodes. In your case, as you are not putting constraints on the localization of the optodes on the head, you should not select "FluenceRegion" and "FluenceExclude" as a head scout but check the "Use default search space" button.
here when selecting FluenceRegion / FluenceExclude, uou are selecting FluenceRegion meaning you are saying you can put optodes anywhere on the head. Using FluenceRegion as a region would work too but is way too computationally expensive as you would need to compute the sensitivity to the ROI for all vertex of the head when you know that the montage should be "above" the ROI. That's why when selecting default search space, you will only look for a region 4 or 5cm away from the ROI.
For future reference:
FluenceRegion is here to describe the region on the head where you can put optodes. When creating the default search space, nirstorm is taking the intersection between the created region and the FluenceRegion.
FluenceExclude describe the region where you cannot put optodes and can be used for three main purposes: Exclude region near the eyes / hear; Exclude region where you know you already have optodes or where you want to put EEG, and remove the part where you cannot compute fluences for example if there is an issue with the head mesh due to noisy MRI. When creating the default search space, nirstorm is taking the difference between the created region and the ExcludeRegion.
Thank you very much Edouard, that indeed helps a lot. I will update my process accordingly and see what I can get out of it. I understand your explanation why the tool may not be appropriate for spatially wide ROIs, and about the FluenceRegion too.
Happy to help with the online documentation, how would you like me to send over material (text, screenshots)?
