Measuring Brain Waves

Inhibition EEG head set

Here are some insights from a 2-day workshop building a headset for measuring brain activity using Electroencephalography known as EEG.

I clearly have barely touched the surface of EEG measurement but here are some key aspects I learned that made this topic more accessible to me.

Measure

Measuring brain activity is about reading tiny voltages

In fact they are so tiny that the main challenge is to isolate them from general everyday interference caused by electricty and mobile radios. Error inducing noise is everywhere. Consider a trip to the woods.

Soldering the board where the probe will be attached. This is also where the signals are amplified and filtered

The art of ballancing amplification and filtering

To get meaningful EEG readings we need to amplify our tiny brain wave voltages a lot. The dilemma is that this also amplifies all the unwanted environmental noise. To isolate valuable parts and cut out noise you apply filters. Finding the sweet spot between amplification and applying the right filters that don’t obstruct the actual data you are interested in is key. It’s tricky.

EEG readings can be saved as simple .wav files

The core signal of one probe is a single analog value over time. An easy way to save these values once digitized is using a simple .wav file. This also allows you to use normal audio software for processing like normalisation for doing some basic analysis and pattern spotting. Handy.

A random 10 seconds sample

Correlate

From signal to meaning

You might find yourself looking for a single EEG reading that directly reflects e.g. your stress level at a given moment in time. Unfortunately there’s no such direct correlation. Instead, you need to look at dominant frequencies and the distribution of multiple frequency spectrums over a period of time. Neural Networks are here to help you to with this otherwise math-heavy task.

Assembled head set with improvised head mount

Summary

Measuring brain waves is a fascinating field and being able to extract data that actually correlates with your perceived mental activity feels extremely rewarding.

While debugging your headset can be difficult it’s not as hard and mysterious to get started as you might think it is. A C.H.I.P. micro controller, some Sugru, probes, gel and some standard electronics parts and you are ready to join the party.

To go deeper you can find a complete set of instructions and schematics on the Inhibition project page.