Back to EEG processing…
It’s the day after thanksgiving and I made my annual I don’t want to have to cook for the next week or two spread of tasty foods. I don’t do the traditional Thanksgiving foods though, I prefer Mexican dishes (since I’m Mexican). Namely I make a huge batch of tamales, since they are a bit of work I tend to make a comically large amount, think 100 or more. So while I have food to last, I need to get back to my main focus and that’s work!
Yes, I ate a lot of food. I also cooked enough for a small army, but had no one to feed. I do that on purpose though, now my fridge is stocked with things I can just warm and eat without having to do anything special. It’s a nice way to save time even though I spent two days making it all. Unfortunately it’s a lot of work up front so now I’m feeling worn out even though I’ve got to get some of my project done.
For those just joining me, this is awkward, I’ve spent most of this post talking food, but I did an experiment! This experiment came with a years worth of funding and in return I would do the experiment, analysis, and publish the results. I’m working on all of that now, but I’m on a deadline and due to COVID the experiment got pushed way, WAY back.
We had just started it when COVID hit, so what would that be, like May? June? I don’t even know anymore, time is useless. The point is that I needed a lot of time to do the analysis, which was due at the beginning of December. Well I just finished collecting the data, as in just a few weeks ago so instead of months, I get weeks to clean the data, perform the analysis, and interpret the results.
I’ve been making a lot of progress, so that isn’t a problem. I’ve got a good amount of the work done so now I’m focusing on getting things organized. There is one thing stressing me out though, I need to do a functional connectivity analysis and I’ve never actually done one before.
Here’s a quick rundown of what I have left. The data is ready to be looked at so now I need to use granger causality to determine the flow of data from one part of the brain to another. What’s granger causality? Well in short, you give it two strings of data taken at the same time so in my case, two EEG electrodes from the times I’m interested in. If using the second set of data helps predict the first set of data BETTER than the first dataset, we say that that the second sensor influenced the first (causality isn’t a good descriptor for the technique even though that’s what we call it).
Basically it helps us estimate the flow of information in the brain, we can look at the flow in both directions IE sensor 1 —> sensor 2 and sensor 2 —-> sensor 1 to see which way (if any) information is flowing. Say we gave the algorithm two copies of sensor 1, well since the second set of data (sensor 1 copy) doesn’t predict it BETTER than sensor 1 alone, we say there is no influence.
That short example is why this works and why we don’t have to worry about neighboring sensors recording the same data from a bridge or some other error. Since I used a word, I should define a bridge. A sensor bridge just means we put too much of the conductive gel on the sensor. We use a large gauge blunt tipped needle to add gel between the scalp and the sensor, we don’t stab so no breaking skin! Since the sensors are so close together, if we put too much gel, the gel can bridge the gap between sensors so they will “share” information with each other. In essence, now the two sensors are copies of one another, not exactly, but close enough that there are problems for us if we don’t notice it after the experiment.
All this to say, I have less than a week to get the analysis done, done right, and get some conclusions drawn. Most of the time when I’m doing research, it isn’t this stressful. There are cases like this for example where I’m up against a deadline and my PI NEEDS me to get it done by that time no matter how many problems we’ve run into.
Back to work I go!