'Review' on nirstorm

[GuyRens]

Dear developers,

I have been exploring nirstorm for quite some time now. Last Friday, I presented the toolbox at the journal club of our research group and I want to provide you with the general opinion of our group as constructive feedback. Although we are unsure whether we will continue with nirstorm, I hope this feedback can help you regardless of it.

Pros:

1. The GUI is intuitive and provides very efficient pipeline editing.
2. The visualizations might be the best so far. We are unaware if other fNIRS toolboxes can do cortical reconstruction ('solving the inverse problem')
3. your optimal montage design is an incredible useful tool for research groups like ours which are interested in topography

Cons:

1. Data processing seem to be limited. Even though detrending and bandpass filters are possible, the field is slowly moving towards more complex techniques such as ICA or PCA.
2. The toolbox can sometimes feel counter-intuitive . For instance, it possible to see the montage on top of the head (although it is not possible to show the brain below the montage/channels). Another example: It is not possible to select channels of interest and only presenting them on the brain (akin to showing the sensitivity model for specific channels)

It is possible that some of the cons I mention are caused by (a) me having missed some things or (b) the relatively young age of the toolbox. Regardless of this, I hope this feedback can be useful for this wonderful project. Please let me know if you have any questions.

Best wishes

Guy

[Edouard2laire]

Hello Guy !
Thanks for your very insightful comment on NIRSTORM.

Few comments: first, it's important to note that, in contrast to other software such as Brainstorm, NIRSTORM is mainly maintained by students of our lab. So the development is mainly driven by own research needs as NIRSTORM is mainly a side project. Currently, it's mainly only me being active on nirstorm, and would greatly appreciate additional help that could take different forms such as help for writing the tutorial or integrating new functions.

We had a meeting, recently, with Dr. Emma Duerden. I think she is from your lab, and we were talking about maybe collaborating or organizing some courses on nistorm. Do you have more information regarding that matter?

Now, concerning, your particular comments:

Data processing seem to be limited. Even though detrending and bandpass filters are possible, the field is slowly moving towards more complex techniques such as ICA or PCA.

yes, it's something we have to study. I know that it's possible to perform ICA and PCA in brainstorm for EEG/MEG. We should review with them the adjuster that has to be made in the code to allow the user to use it for nirs.

The toolbox can sometimes feel counter-intuitive . For instance, it possible to see the montage on top of the head (although it is not possible to show the brain below the montage/channels). Another example: It is not possible to select channels of interest and only presenting them on the brain (akin to showing the sensitivity model for specific channels)

yes, I agree. The main issue here is the lack of complete proper tutorials. In fact, akin to brainstorm, the main way to learn how to use nirstorm is to have a local expert that can teach you the different tricks. (Even if Brainstorm tutorial are very well written, most of my knowledge of the software comes from local mini-course, or discussion with more experimented users)

The thing is I found that written tutorial is not very efficient to teach how to use GUI software as:
From our perspective, they are very long to write as each step and click can have its importance (especially for displaying the result).
From the user perspective, they are not enough to learn to fully learn how to use the software as they do not teach those small tricks that are very useful daily for display.

We have two options here; either spending time on writing very complete tutorials or what I was thinking recently, is to have video tutorials. We may do some webinar on nirstorm soon; so if it's working well, I think I might suggest we generalize the idea for all the tutorials. What do you think of that idea?

Finally, I started a todo list of what should be done in nirstorm here: https://github.com/Nirstorm/nirstorm/wiki/What's-next-%3F . Feel free to expend it with new ideas :)

Best regards,
Edouard

[GuyRens]

Hi Edouard,

Sorry for my very delayed response.
Over the past few months we have looking into some other toolboxes.
We are likely to use nirs-toolbox (Ted Huppert) for our data analyses.
However, we noticed that his forward modelling and cortical reconstruction seem to be limited and undermaintained. I believe he will keep his focus on analysis methods rather than visualization.

We still believe that the visualization of nirstorm is still one of the better around (although we haven't had a detailed look into other options).

We may be interested in using the visualizations of nirstorm for our papers in the (near-) future. I was wondering whether you are interested in having a video-call to explore our options? If you are interested, please provide me with your contact information and we could continue this discussion over e-mail.
If we would go with nirstorm, we could provide support with our research group in the further development.

Best wishes

Guy

[Edouard2laire]

hello Guy,

Again thanks a lot for your interest in nirstorm. We would indeed be very interested in collaborating in the further development of nirstorm. You can contact me using: edouard.delaire@concordia.ca I don't think we will have enough time to schedule a meeting before the winter break, but we can already discuss about a meeting in early

Over the past few months we have looking into some other toolboxes.
We are likely to use nirs-toolbox (Ted Huppert) for our data analyses.
However, we noticed that his forward modelling and cortical reconstruction seem to be limited and undermaintained. I believe he will keep his focus on analysis methods rather than visualization.

I agree with you. One solution that we can discuss would be to write a wrapper for the corresponding function of the nirs-toolbox you want to use in nirstorm. The idea would be to be able to directly call those functions from nirstorm and save the result in nirstorm thus benefiting of both the rich data analysis functions of the nirs-toolbox and the visualization of nirstorm. This is something we obviously would need to discuss with Dr Huppert but that should not be a problem if the toolbox is properly acknowledged. (i know for example that you can call process from Homer2 directly from the nirs-toolbox so such inter-toolbox communication already exist)

Best regards,
Edouard