Re:fMRI measures blood flow, not brain activity (Score 2, Informative)

Anonymous Coward said:

Less than 50% of the time, when you average the neural activity over several SECONDS (an action potential lasts 0.015s), and over 1 cubic CENTIMETER (containing 10^8 neurons), fMRI tells you something about that average activity.

While this is true for the majority of fMRI work done today, things are changing. Higher field strengths have greatly increased the spatial resolution of fMRI. Typical voxel size at 3 Tesla is down to about 3 mm to the side (echoplaner, BOLD contrast). As higher field strengths become more common, voxels will become much smaller.

Temporal resolution has also improved. A lot of the older fMRI work uses block designs (e.g., two conditions, control and experimental, 30 second each blocks). This reliance on averaged data has kept temporal resolution low, and has been an impediment to observing the sequence of activations in the brain in response to stimuli. But recently, several labs have reported good results with event-related designs, which can show the sequence of activations in response to a stimulus. This information is much more useful in understanding networks in the brain, rather than just seeing time-averaged regions that "light up". There are also methods available now (still experimental) which can detect the BOLD response in a single activation. No averaging required.

Higher field strengths also allow the use of nuclei other than hydrogen in fMRI imaging. Typically, sodium imaging is done with a repetition time of 100 milliseconds. Higher field strengths are also useful in hydrogen imaging, where you can image the brain faster (or image a larger volume of brain) than before.

I do agree with the previous posts that point out the huge gap in understanding between observing an fMRI map, no matter how good, and coming to conclusions that could materially help an advertising campaign. We just don't know enough about the brain to be able to image thoughts unambiguously.

I think progress in this field will be rapid. More powerful scanners are coming online at many institutions across the US. There is now a 9.4 Tesla full-body scanner at the University of Illinois. This should allow very fast acquisitions, and also the imaging of metabolically significant nuclei other than hydrogen. fMRI currently looks for regional increases in blood flow which accompany activations, but the ability to image other nuclei will allow fMRI to get much closer to what is happening at the level of the neuron.