fMRI measures BOLD, the Blood Oxygenation Level Dependent response. As the name says, BOLD is when a bit of the brain becomes more active, it uses more oxygen, and the oxygenation level of the blood in the area drops - although it then increases to compensate, and it's the increase that most fMRI picks up.
There's a catch though: blood flows. Specifically, it flows from arteries, into tissues - the brain, in this case - and then into veins. Blood leaving the brain tends to end up in the larger veins and, being large, these exert a large effect on BOLD - even though they're some distance from the true site of neural activation.
So, the worry is that BOLD blobs may be shifted towards the nearest large vein, reducing the accuracy of fMRI. It's a well-recognized issue, but it's not clear just how serious it is... or what we can do about it.
Enter Canadian researchers Ismael Gaxiola-Valdez and Bradley Goodyear. By comparing BOLD to the alternative method of ASL - They show that the large vessel effect does happen to BOLD signals, shifting the activation blobs nearer to the surface of the brain (where veins are.) However, the effect is more pronounced when the blobs represent absolute BOLD changes. When blobs represent the z score - which they usually do - the difference is smaller.
That's somewhat reassuring... but here's the bad news. Some peoples' brains show larger BOLD changes than others. Gaxiola-Valdez and Goodyear show that these individual differences in BOLD magnitude are probably driven by large vessel effects because as the pic at the top shows, BOLD was most variable (between people) at the surface, i.e. near the veins, while ASL, less subject to the vein bias, was most variable deep in the brain, as you'd expect.
Overall, it's an important paper and a reminder that, although neuroscientists sometimes treat the brain as separate from the rest of the body, it's not.
Gaxiola-Valdez I, and Goodyear BG (2012). Origins of intersubject variability of blood oxygenation level dependent and arterial spin labeling fMRI: implications for quantification of brain activity. Magnetic resonance imaging PMID: 22795932
5 comments:
Yeah well. Just remember who said it first.(a 1000 times) It's a virtual representation wholly generated by software, not a video.
It's a CGI of the brain, somewhat related to actual goings on.
Why won't nobody recognize my genius ;) Sheesh, how hard can it be to just type: Yes Petrossa you were right all along.
From the article:
Pre-processing included the following separate processing steps: "motion correction using MCFLIRT, interleaved slice timing correction," the same preprocessing steps were applied to both the BOLD and ASL data. Given that these two processing steps can not be separated but are and given that the temporal dynamics of ASL and BOLD are expected to be different I don't know what to make of this paper. I see at as one more case of story telling in neuroscience.
This seems like a completely bogus comparison to me. The difference in correlation is probably driven by the fact that the sensitivity will be much greater for BOLD with the long blocks compared to the short blocks (because of the low efficiency of the 6 secs on/12 secs off paradigm). If the signal is weaker (and the noise thus higher) then of course there will be a lower correlation!
Russ raises an excellent point here. On that note, are the two correlations significantly different?
Another thing that struck me is that this doesn't really suggest anything. Even if individual or group differences measured with BOLD are really hemodynamic, they must originate somewhere. This is of course why cognitive neuroscience is all about combining multimodal evidence. It doesn't make sense to conclude anything from a single fMRI study - but if several studies, TMS, lesions, pharmacology, and animal models all point to similar findings one can be more confident they might be meaningful.
Post a Comment