tag:blogger.com,1999:blog-4402160631955197288.post1460572458582871876..comments2024-03-12T19:57:17.818-07:00Comments on practiCal fMRI: the nuts & bolts: Resting state fMRI: is there an optimal protocol?practiCal fMRIhttp://www.blogger.com/profile/07387300671699742416noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-4402160631955197288.post-2043774756763932752013-01-27T18:35:02.323-08:002013-01-27T18:35:02.323-08:00Hi Mayur,
You ask some pertinent questions. I don...Hi Mayur,<br /><br />You ask some pertinent questions. I don't think the 2D vs 3D comparisons are easy to do, or to evaluate. What can we say in general? Well, any use of reference scans for acceleration, such as the ACS for GRAPPA, is going to enhance motion sensitivity. Assuming comparable schemes for acceleration, e.g. multiband, for 2D EPI vs (3D) EVI, so that the systematic effects of prescan to time series mismatch can be assumed to be comparable, then one might expect the enhanced SNR of the 3D scan, along with the faster TR, to offer some benefits over the 2D EPI. But there may be another cost, e.g. increased ghosting in the 2nd phase encoding dimension. <br /><br />I'm going to be tracking the "accelerated" fMRI literature closely so as people present 3D methods I'll try to review them here. practiCal fMRIhttps://www.blogger.com/profile/07387300671699742416noreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-46628731697743444652013-01-25T07:09:32.933-08:002013-01-25T07:09:32.933-08:00Going for subsecond (or less than 500ms) volume TR...Going for subsecond (or less than 500ms) volume TR can make physiological noise characterization better for the ICA analysis. With segmented 3DEPI which inherently provides you better SNR than multislice 2DEPI, how big does the SNR dropout problem remain? Additional advantage of using 3D instead of 2D is to gain higher reduction in volume TR due to its ability to accelerate in the second phase encode direction. Recent developments, especially CAIPIRINHA, can help mitigate some of this SNR loss due to use of GRAPPA.<br />I see that you have 7 posts related to resting state fMRI and I am replying just after reading the first one. But I couldn't stop myself from jotting down thoughts which 'sparked in my mind' immediately after reading the post. I shall of course go through all the posts in time.<br /><br />Very nice work!<br /><br />- MayurMayur Narsudehttps://www.blogger.com/profile/09602964497886301352noreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-27106916750535134722011-01-13T14:35:14.535-08:002011-01-13T14:35:14.535-08:00@Anonymous (again):
I just re-read your comment s...@Anonymous (again):<br /><br />I just re-read your comment slowly... I think your question actually concerns what is generally called "functional connectivity," as performed with a seeded correlation analysis in 1995 by Biswal et al. If the time series you are using came from an explicit task then it won't, by current definitions, be "resting state." In my comment of an hour ago I was really getting to the heart of what it means to be "resting" while a time series acquisition is happening. <br /><br />Sorry if my previous comment confused the issue there!practiCal fMRIhttps://www.blogger.com/profile/07387300671699742416noreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-55350801802589775462011-01-13T12:57:08.617-08:002011-01-13T12:57:08.617-08:00@Anonymous:
This is a timely question! I had been...@Anonymous:<br /><br />This is a timely question! I had been thinking about the limits of resting state for a while, even drafting a post on it. I'm still thinking about my post. However, for now there are two pertinent papers to consider:<br /><br />Waites et al. (Hum Brain Mapp. 2005 Jan;24(1):59-68) showed that prior task effects can "hang over" into a 'resting' state scan.<br /><br />Benjamin et al. (Front Hum Neurosci. 2010 Dec 1;4:218.) showed just last month that the specific instructions given to a subject on whether to actively ignore the scanner sounds or just rest also had a profound effect on the networks produced.<br /><br />Together with the van Dijk result that eyes open rest with or without target are similar, whereas eyes closed is dissimilar, then we can take it that (1) it may be important to consider doing the rs-fMRI at the start of a session if the subsequent stimuli in any task might produce hangovers, and (2) that we should tell the subject nothing more than to lie still, eyes open and rest comfortably.<br /><br />I'm still thinking about my draft post on the limits of rs-fMRI. But at the very least I will do a more detailed post on the Benjamin paper soon.practiCal fMRIhttps://www.blogger.com/profile/07387300671699742416noreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-11409527996610297732011-01-13T07:04:39.571-08:002011-01-13T07:04:39.571-08:00Hi, thank you for this very informative post! This...Hi, thank you for this very informative post! This is really useful.<br />I have a question that you might be able to answer:<br />I am really naiive when it comes to rsfMRI, but I was always wondering whether you could just do the rsfMRI analysis on a 'normal' task based dataset? I mean it should make such a difference to the overall activation or does it? Do you know whether this has been done?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-2445004001010209082010-11-21T04:02:15.375-08:002010-11-21T04:02:15.375-08:00Good post and it looks like an important paper. I ...Good post and it looks like an important paper. I didn't realize resting-state scans were so quick: 5 minutes. <a href="http://www.umich.edu/~fmri/asl.html" rel="nofollow">Arterial Spin Labelling (ASL)</a> scans are now down to about 5 minutes too.<br /><br />That's dangerous in some ways because it encourages "...and just one more thing" syndrome - adding more and more 5 minute scans until your experiment is much longer than planned and you regularly over-run your scan slots... and you end up having to analyze a lot of data in a hurry.Neuroskeptichttps://www.blogger.com/profile/06647064768789308157noreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-72321996848207163762010-11-02T20:38:27.810-07:002010-11-02T20:38:27.810-07:00Thanks Jim, very useful. That paper nicely confirm...Thanks Jim, very useful. That paper nicely confirms the risk of global signal regression for anti-correlations. Now to determine a robust alternative that might preserve neurally driven anti-correlated networks!<br /><br />There are several papers that investigate the specific effects of respiration and cardiac fluctuations on resting state signals, but a recent paper from Catie Chang and Gary Glover (Neuroimage, 2009. 47(4): 1381–1393. "Relationship between respiration, end-tidal CO2, and BOLD signals in resting-state fMRI") provokes some more thought about an issue I mentioned in the first post: one wonders how much of a role there is for refluxed CO2 in the average magnet bore. I think I'll try acquiring some of my test resting state data (the data I said I'd get and post) with and without bore ventilation. Then if someone has a mind to, they can investigate whether magnet ventilation is of benefit to interpreting functional connectivity.<br /><br />PS I would have already got test resting state data this week but the scanner has spiking issues that are being addressed. Look for the data within the next week.practiCal fMRIhttps://www.blogger.com/profile/07387300671699742416noreply@blogger.comtag:blogger.com,1999:blog-4402160631955197288.post-60275868112235876492010-11-02T12:50:54.777-07:002010-11-02T12:50:54.777-07:00If you haven't seen this article yet, it also ...If you haven't seen this article yet, it also seems to have some useful info on processing steps that make a difference.<br /><br />Correlations and anticorrelations in resting-state functional connectivity MRI: A quantitative comparison of preprocessing strategies.<br /><br />Andreas Weissenbacher, Christian Kasess, Florian Gerstl, Rupert Lanzenberger, Ewald Moser, Christian Windischberger.<br /><br />NeuroImage 47 (2009) 1408-1416Jim Eliassenhttps://www.blogger.com/profile/10760918895137802811noreply@blogger.com