[FieldTrip] epilepsy example

[Schoffelen, J.M. (Jan Mathijs)]

Cher Alain,

Thanks for reporting this! The SQUID jump indeed is quite prominent. In the absence of the ground truth, I quickly computed a volumetric image of kurtosis, using data from which the jump was cut out (prior to filtering). Interestingly, the spatial image does not look altogether too different (apart from the scale) between the two situations (left panels in the attached screenshots always refer to the ‘original’ results, right panels to the results obtained with the jump cut out). These spatial images were obtained with covariance matrices that obviously were quite different. I think that the beamformer to a large extent protects against the disruptive effect of the SQUID jump (at least spatially) because this jump does not localize well in brain space. On the other hand, it had an effect on the reconstructed kurtosis value throughout the volume.

However, I agree that the data that is subjected to any downstream analysis should be as clean as possible. I will make some small cosmetic changes to the tutorial to do justice to this.

With respect to any ground truth data:

In the context of the 2022 BIOMAG epilepsy data analysis competition I adjusted the kurtosis pipeline to be applied to the competition data. During the conference we discussed the possibility to make the data (+ prize winning analysis pipelines, among which the kurtosis pipeline was one, whoohoo!) publicly available in some way. Despite the initial enthusiasm (of the competitors + organizers(: Jean-Michel Badier and Christian Bénar)) this ended up on everybody’s backburner unfortunately. Yet, I am pretty sure that the competition data have been labeled.

Best wishes,
Jan-Mathijs


[cid:47D17A13-8703-4B3C-86CA-D31893331724]
[cid:AE32304A-8001-4A10-AA48-876F890F58D5]



On 24 Mar 2024, at 19:00, Alain de Cheveigne via fieldtrip <fieldtrip at science.ru.nl<mailto:fieldtrip at science.ru.nl>> wrote:

Hi all,

I've been going through the 'Virtual channel analysis of epilepsy MEG data' tutorial (https://www.fieldtriptoolbox.org/tutorial/epilepsy/<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.fieldtriptoolbox.org%2Ftutorial%2Fepilepsy%2F&data=05%7C02%7Cfieldtrip%40science.ru.nl%7C2f712284a8a341bec31c08dc4d843a59%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638470477509485173%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=OlOcYeraMNfryIIj7tCBAVzcz3zeudR0bZX%2BJtC0YDo%3D&reserved=0>), and I noticed that the CTF data for case3 has what looks like a flux jump, see below.  The code applies a high-pass filter that turns this into a spike which might drive the locations of high kurtosis in the beamformer images.

Other examples seem OK, but it's hard to judge the outcome without ground truth.

Is anyone aware of MEG/MRI data with labels that could be plugged into the tutorial?  The ideal would be data labeled for the type of activity that a surgeon would really like to localize, e.g. interictal or pre-ictal onset.

Alain

_______________________________________________
fieldtrip mailing list
https://mailman.science.ru.nl/mailman/listinfo/fieldtrip<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fmailman.science.ru.nl%2Fmailman%2Flistinfo%2Ffieldtrip&data=05%7C02%7Cfieldtrip%40science.ru.nl%7C2f712284a8a341bec31c08dc4d843a59%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638470477509485173%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=jM9jd5Kta%2B3%2BOWZsG2Yj6g3AEgRb2Fh77W%2BZixVFhWc%3D&reserved=0>
https://doi.org/10.1371/journal.pcbi.1002202<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1371%2Fjournal.pcbi.1002202&data=05%7C02%7Cfieldtrip%40science.ru.nl%7C2f712284a8a341bec31c08dc4d843a59%7C084578d9400d4a5aa7c7e76ca47af400%7C1%7C0%7C638470477509641411%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=FkUj3VulsFNiOg6R9XJ3dU62ilAlZLoxZCW5%2BzZvrNM%3D&reserved=0>
<PastedGraphic-2.pdf>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20240326/2475acbe/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screenshot 2024-03-26 at 11.41.55.png
Type: image/png
Size: 116022 bytes
Desc: Screenshot 2024-03-26 at 11.41.55.png
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20240326/2475acbe/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screenshot 2024-03-26 at 11.31.06.png
Type: image/png
Size: 172386 bytes
Desc: Screenshot 2024-03-26 at 11.31.06.png
URL: <http://mailman.science.ru.nl/pipermail/fieldtrip/attachments/20240326/2475acbe/attachment-0003.png>