Hacker News Reader

Comment by mrcrm9494 | Hacker News Reader

this headline is a bit misleading on the first read, since it only affects functional (f)MRI, which is controversial since a longer time. a prominent example is the activity that has been detected in a dead salmon

View on HN

The dead salmon was just a lesson in failing to correct for multiple comparisons.

As the first author of the salmon paper, yes, this was exactly our point. fMRI can be an amazing tool, but if you are going to trust the results you need to have proper statistical corrections along the way.

Cheers!

It's not that fMRI itself is controversial, it's that it is prone to statistical abuse unless you're careful in how you analyse the data. That's what the dead salmon study showed - some voxels will appear "active" purely by statistical chance, so without correction you will get spurious activations.

This study questions the fMRI method itself, not the statistical analysis (you're right that the dead salmon study was challenging the way statistical analysis is done). Basically, this study claims that the association between the BOLD signal measured by fMRI and actual brain activity is quite weak, and they are even anti-correlated in 40% of cases.

There is no statistical analysis that can save you if your interpretation of a signal is wrong (for example, you can't get information about personality from phrenology, regardless of what statistical analysis you try to apply to the data). That's not to say that we need to just trust this study implicitly - I'm just trying to describe how serious of a problem to the field their claim is.

So, is fMRI like "fast" MRI? Can someone fill the rest of us mortals in on this? :)

f is functional. MRIs are basically huge magnets used for imaging. When you apply a strong magnetic field, different tissue types and densities will react differently, and the MRI is basically measuring how those tissues react to the magnet. It is very good for imaging soft tissues, but not so much bone. Someone figured out that you can measure blood flow using the MRI, because blood cells react in a magnetic field, then "relax" at a known rate. Since we can measure blood flow, that is correlated with increased brain activity, i.e. since more neurons are firing, they require more energy, and therefore more blood. So, fMRI is using blood flow as a proxy for brain activity.

Fmri doesn’t measure blood flow, it measures the oxygen level in the blood. Hemoglobin molecules change shape when they carry oxygen and the different shapes react differently to magnets, which is a real stroke of luck

Yep, this is why it's also called BOLD imaging, for blood-oxygenation-level-dependent fMRI. I did my PhD is BME and brain-computer interfaces, but it has been a while since I worked in the field.

it doesn't measure the oxygen level directly either. the bold signal is correlated to dephasing induced by the oxy/deoxy hg ratio that isn't even necessarially localized to the voxel (flow or long range magnetic susceptibility perturbations from nearby accumulated deoxyhg (veins)).

I'm going to follow on a bit from what jawilson said. The idea is this - you can measure blood oxygenation by sticking your head in a big magnet that makes atoms spin really fast and measuring the radio waves that come off. This is imprecise, but reasonably repeatable.

So if I show you a picture of a cat, and you like cats, then a bit of your brain might start using more oxygen because you're thinking about cute furry things, and if I show you a picture of a car, and you like cars, a different bit of your brain lights up showing more oxygen use because you're thinking about fast shiny things.

But really we've only got the barest idea of what bits of the brain do what, and maybe it's a bit of brain that goes "hey I'm happy" that lights up in both cases because you like both cats and cars.

We can kind of see bits we think are associated with muscle movement coming to life if I show you a picture of a bike, and you like cycling, and if I show you a really cool mountain track you imagine belting down it flat out. That lights up differently if I show you something else.

However, we do not really know except in very broad terms what bits of the brain actually do what. We can't "see thoughts", we just know that some bits of brain seem to use more oxygen than others, and from that we guess "this bit of brain is for thinking about sitting in a nice cafe with a cup of coffee and a newspaper" versus "this bit of brain is for being frightened of lions".

At least when phrenology was a thing, the ceramic heads with lines painted on were inexpensive and didn't require three-phase power and huge barrels of liquid helium.

If you apply enough gain and filtering to an unknown signal, eventually you'll pull something out of it that you can convince yourself is what you're looking for.

wondering why you are downvoted. You are right, though it's kind of inferred that the author means fMRI as the title focuses on brain activity only.

Structural MRI does not record brain activity, because it is, like, structural, not functional.

Structural MRI is even more abused, where people find "differences" between 2 groups with ridiculously small sample sizes.

Reply on news.ycombinator.com