[Eeglablist] Network Based Statistics (NBS) thresholds

[Bahman Nasseroleslami]

Dear all,

Just wanted to mention that for high-dimensional statistics, an alternative
approach is to use Empirical Bayesian Inference (developed by Efron, 2007,
doi.org/10.1214/009053606000001460). Our group have used it successfully in
our 2 recent publications (doi.org/10.1093/cercor/bhx301 and
doi.org/10.3389/fneur.2017.00395) and an implementation of EBI for MATLAB
is available on GitHub (github.com/NeuroMotor-org/EBI) with a preprint on
bioRxiv (doi.org/10.1101/342964). Sorry that part of the methods are from
our own group.

Best
Bahman

––––––––––––
Bahman Nasseroleslami, PhD
Senior Research Fellow
Academic Unit of Neurology, School of Medicine
Trinity College Dublin, the University of Dublin
Dublin, Ireland.

Room 5.43, Trinity Biomedical Sciences Institute
152-160 Pearse Street, Dublin D02 R590, Ireland.
nasserob at tcd.ie
www.tcd.ie/medicine/staff/nasserob/
––––––––––––

On 1 August 2018 at 02:15, Makoto Miyakoshi <mmiyakoshi at ucsd.edu> wrote:

> Thanks Luca for your prompt response.
> Ramtin, it's a good chance to ask anything to Luca :-)
>
> Makoto
>
> On Tue, Jul 31, 2018 at 4:17 PM Luca Cocchi <Luca.Cocchi at qimrberghofer.
> edu.au> wrote:
>
>> Hi Makoto, I trust you are well. Let’s see if I can help:
>>
>>
>>
>> The initial threshold (F or T values) should be defined based on the
>> expected effect size of the effect of interest. It is not a good practice
>> to choose the threshold arbitrarily or explore a large range of possible
>> thresholds. Technically, exploring a range of thresholds introduces an
>> additional multiple comparison problem. In its essence, the approach used
>> in the NBS (initial thresholding and subsequent cluster statistic on
>> surviving edges) is not different from what is done in other popular
>> analysis packages (e.g., SPM, recommended search threshold of p<0.001).
>>
>>
>>
>> In the field called “Significance” the user can define the desired NBS
>> (network-wise FWE corrected) threshold. Typically, this threshold is set to
>> 0.05. Just to be clear, this is not the initial “search” threshold but the
>> threshold ascribed to define a significant network at the given value. Note
>> that significance is ascribed at the level of the whole network and not
>> individual edges.
>>
>>
>>
>> Further information can be found in the NBS manual and the related forum.
>>
>>
>>
>> Cheers, Luca
>>
>>
>>
>>
>>
>> *From:* Makoto Miyakoshi <mmiyakoshi at ucsd.edu>
>> *Sent:* Wednesday, August 1, 2018 3:26 AM
>> *To:* R.Mehraram2 at newcastle.ac.uk; Luca Cocchi <
>> Luca.Cocchi at qimrberghofer.edu.au>; andrewzalesky at gmail.com
>> *Cc:* EEGLAB List <eeglablist at sccn.ucsd.edu>
>> *Subject:* Re: [Eeglablist] Network Based Statistics (NBS) thresholds
>>
>>
>>
>> Dear Andrew and Luca,
>>
>>
>>
>> It's been a while since we met in Tokyo. How are you doing? I wonder if
>> you have time to kindly help Ramtin.
>>
>>
>>
>> Makoto
>>
>>
>>
>> On Thu, Jul 26, 2018 at 2:33 AM Ramtin Mehraram (Student) <
>> R.Mehraram2 at newcastle.ac.uk> wrote:
>>
>> Hi all,
>>
>>
>>
>> I am struggling in understanding what the threshold values in the NBS
>> toolbox mean. I hope you can help me.
>>
>>
>>
>> To my knowledge, the value which in the toolbox GUI is defined as
>> “Threshold” should be arbitrarily chosen, but I am not sure in which way
>> that value influences the output. Is it a threshold on the t values?
>>
>>
>>
>> On the bottom, there is a field called “Significance”, which is
>> supposedly the max p-value to be considered significant. However I change
>> that value, the output seems not to be affected at all, showing always
>> p<0.001 in the viewer window.
>>
>>
>>
>> Am I missing something?
>>
>>
>>
>> Thank you in advance for your help.
>>
>>
>>
>> B.
>>
>>
>>
>> Ramtin Mehraram
>>
>> PhD Student @ramtinTVT
>>
>> Biomedical Research Building 3rd floor
>>
>> Institute of Neuroscience
>>
>> Newcastle University
>>
>> NE4 5PL, United Kingdom
>>
>> www.lewybodylab.org
>>
>> https://www.newcastlebrc.nihr.ac.uk/research-themes/dementia/
>>
>>
>>
>> _______________________________________________
>> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
>> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.
>> ucsd.edu
>> For digest mode, send an email with the subject "set digest mime" to
>> eeglablist-request at sccn.ucsd.edu
>>
>>
>>
>>
>> --
>>
>> Makoto Miyakoshi
>> Swartz Center for Computational Neuroscience
>> Institute for Neural Computation, University of California San Diego
>>
>
>
> --
> Makoto Miyakoshi
> Swartz Center for Computational Neuroscience
> Institute for Neural Computation, University of California San Diego
>
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to eeglablist-unsubscribe at sccn.
> ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20180802/7bc0dd5a/attachment.html>