Acquiring a simple EEG signal

Here is an example of using the component software to acquire a simple occipital alpha band EEG signal. In the image below, the topmost plots show 1 second of the current input signal (although this particular record contains more EMG than EEG), the 0-40 Hz spectrum of the 100 most recent 1 second input records as a cascade plot (the current spectrum is at the front of the plot), and an average of the sprectra of all 100 inputs showing a peak at the alpha frequency (9.88 Hz in this case).

(Click for larger image)

Here is the component software "program" which acquires and measures this data:

In this example, two first-in-first-out (FIFO) queues save 100 seconds of both the input signal and the computed spectrum records. A statistics component calculates the moving average of all computed spectra in real time.

This data was acquired using amplifier 5. Two electrodes made from 1964 FDR silver dimes attached to 24 gauge wire were applied wet using conductive "goop" (Neuroscan Quickgel) to positions O2 and P4 to acquire a bipolar signal from the right occipital area. 100 seconds of data acquired at 256 samples/sec were recorded with the eyes closed in a darkened room. The AC power was turned off during data acquisition so that no 60 Hz (120, 180, etc...) line noise was present.

This is an example of acquiring "bipolar" EEG data, since both electrodes are connected to specific locations over the scalp. Here is a map of the standard electrode locations used to acquire EEG data. These locations actually exist in 3d spherical coordinates rather than in 2d, and correspond to the electrode positions of my Neuroscan Quickcap electrode cap (although my cap has only 32 channels, so that it corresponds to only a subset of the positions shown below). However, in this example two individual electrodes were used rather than the cap.

Here is a subset of the 100 seconds of data acquired. Each plot represents 1 second of data, and contains 256 points. The 10 Hz alpha signal component can be clearly seen in many of these records, as can some muscle movement (EMG) at higher frequencies (e.g. in plots 66 and 67). The most recent data is at the top of the window.

Here are details of the multiple spectra and average spectrum for this data from 0-20 Hz at a resolution of 0.5 Hz/point. Again, the alpha peak can clearly be seen, this time exactly at 10 Hz.

Here are 25 seconds out of 30 of another recorded set of occipital EEG data. This data was acquired with the eyes open, watching the computer display in real time.

Although visual inspection does not immediately suggest the presence of an alpha band component in this signal, it is indeed there. Here are spectral records and the average spectrum for this 30 second dataset, from 0-32 Hz, at a resolution of 1 Hz/pt. In this case, the peak of the average spectrum is again at 10 Hz.

In the next example (and most likely in all future examples), EEG data was acquired using a "unipolar" electrode configuration. In this case, only one of the electrodes (that connected to the positive input of the amplifier) is attached to a point over the scalp (in the following, again at O2). The other electrode (providing the return current to the negative input of the amplifier), is attached to a common, or "reference", position. In this case, the return/reference is provided by a loop of wire wrapped around the neck. In unipolar EEG acquisition, all negative electrodes are connected to this same reference electrode (i.e. all channels have as their negative input this common electrode).

Below is a photo of the two electrodes used to perform unipolar EEG acquisition. The positive electrode has been glued to a small peice of Velcro which can be attached to the inside of a matching Velcro headband. This electrode can be placed anywhere on the band, and the band then firmly attached around the head. Conductive goop is then injected between the electrode and scalp using a large diameter hypodermic (in this case "hyper") needle and syringe. Although this arrangement limits electrode positions to those under the headband, it is sufficient to acquire signals from many of the peripheral electrode positions shown in the diagram above. An additional Velcro strap oriented in a longitudinal direction will be added in order to accomodate the remaining electrode positions as required.

The return electrode consists of a length of copper wire mesh (in this case actually 2.2mm wide "Pro Wick" desoldering tape) attached to an aligator clip. The copper mesh is wrapped around the neck 4 times and pulled tight (but not too tight!), and clamped with the clip to form a secure connection. Although this arrangement is sensitive to large whole-head muscle movements, it does seem to be insensitive to small facial movements, speech, and to EKG contamination.

Here is a set of 30 seconds of EEG data acquired using the unipolar electrode configuration described above. This data was acquired with the eyes open, watching the computer display in real time.

Again, although visual inspection does not immediately suggest the presence of an alpha component, it is there. Spectra computed for 0-32 Hz at a resolution of 0.5 Hz/pt show an average peak at 10.5 Hz. This particular dataset will form the basis of the additional spectral analyses shown in the next page.

©Copyright Sky Coyote, 2001-2002.