Knowledge Center

Your Resource for Training and Technical Support
If you need immediate assistance, call us at 1-800-970-6670 or email support@magstimegi.com.

Artifact recognition using high density EEG

EEG is designed to record electrical activity at the head surface in order to assess the underlying electrical activity arising from the brain. EEG is quite sensitive, however, and in addition to recording electrical signals that arise from the brain, EEG will also record activity of non-cerebral origin. Electrical activity arising from sources other than the brain is considered artifact and is divided according to whether it is physiological or extra-physiological. Physiological artifacts are generated from the patient, arising from sources other than the brain, such as cardiac rhythms, eye blinks and muscle movements. Extra-physiological artifacts are generated by sources outside of the patient's body and are instrumental or environmental, such as electrode popping, or electrical mains (50 or 60 Hz) contamination.  In conventional EEG with 19 to 21 channels, artifact recognition is well established and relies on the fact that different types of artifact express consistent characteristics within the data that allow for discernment of the events as artifact.    However, with high density EEG, which EGI considers to be at least 128 channels, electrodes more densely populate the head surface, including areas not covered in conventional EEG, such as below the canthomeatal line. The result of this is that high density EEG changes the way artifact manifests in the data.

This is, of course, important for proper artifact recognition when working with high density EEG. Even more importantly, however, is the fact that a higher density of electrodes describes more of the electrical field across the scalp at any given moment, allowing for more accurate interpretation of events, whether artifact or brain activity.  To illustrate this, we discuss examples of eye blink and eye movement (physiological artifacts) comparing their characteristics in high density EEG vs conventional EEG, and also to demonstrate the level of detail can be gleaned about these events from the high density EEG. 

Eye blinks are recognized as "V" shaped transient potentials in the positive direction that appear in the frontopolar electrodes such as in the image below: 

 01 EB Chart

Image 1: bipolar montage of 10-20 equivalent electrodes; a subset frrom a 256 dense array EEG data set. Polarity is Negative up. Sensitivity is 7.5 µV/mm.

 

We know that when the eye blinks, it creates a positive field above the eye and a negative field below. In conventional EEG, since there are only a few electrodes in the frontopolar region, one can expect the transient positive potential to appear in the Fp1, Fp2 and F7, F8 electrodes. 

However, when we look at this same artifact of all 256 channels plotted out in the (2D) anatomical positions and as though we are looking down from an aerial perspective, we see more of the electrical field described, with both the positive and negative aspects. Note that the Fp1, Fp2, F7 and F8 electrodes are circled for comparison: 

02 EB Plot 

Image 2: Full 256 data set including the 10-20 electrodes as seen in Image 1, but with the data shown in a topo plot view with an aerial perspective. Left=Left and Right=Right. Vertex is found in the middle, back of the head (posterior and occipital channels) are at the bottom of the image. Nasion and cheek electrodes are shown at the top.  Absence of electrodes (holes) seen on the left and right sides are spaces to accommodate the ears during recording.  Polarity is Negative up. Sensitivity is 7.5 µV/mm. Montage is 256 channel average reference. 

Notice the negativity in the sensors in the most anterior left and right positions. These sensors are below the eyes, positioned on the cheeks (cheek placement is essential for measuring the electrical activity of the anterior and basal temporal lobes); therefore these sensors measure the negative gradient of the electrical field that an eye blink generates. Notice also that there is an inversion line where the negative field transitions and the positive field is captured by the many sensors in the frontopolar region, extending as far back as the frontocentral electrodes just anterior to the vertex (FC1 and FC2 in the 10-10 labeling system). 

This shows us the full extent that the electrical potential of an eye blink generates and its propagation across the scalp surface. The high density EEG also validates what we already know to be true, that the source of this potential is indicated by the white line, where the polarity flips between the negative and positive ends of the field and exactly where the eyeballs are located (see Image 3 below). Here is a 3D voltage map illustrating this concept where the “source” of the generator lies right at the center of the dipolar field generated by an eye blink: 

                                                                  03 EB 3D Map

Image 3: Same 256 data set as Image 2, also with an average reference applied, however here the data is shown as a voltage map where red is positive, blue is negative and white is neutral. Saturation of color indicates intensity, i.e. the darker the red, the more positive the value. Map projected onto 3D head where the data (color) in between sensors is interpolated. 

 

To further this point, let’s consider an eye movement artifact. Note that artifact from eye movement will take on different characteristic shapes in the data, depending on how fast the eyeballs are moving. For example, saccades are small, rapid eye movements and have a characteristic square shape in a 10-20 bipolar montage. Here, we consider a slower eye movement (Image 4 below) which can, at first, appear similar to frontal slowing.   

04 EyeM Left Chart

Image 4: Same data file and parameters as Image 1, showing an earlier time point in the file.  Eye movement seen where the yellow line (Time Sync Marker) appears over the data.

 

When we look at the same time point in the full data set, we easily see this is a lateral eye moment, and specifically an eye movement looking to the left: 

05 EM Left

Image 5: Same data file and parameters as Image 2, showing an earlier time point in the file. 

In this plot, we clearly see the positive potential on the left cheek created as the eye looks left, while there is a negative potential on the right. Note that a pure lateral movement would not also create positivity in the left frontopolar electrodes. Therefore the presence of this indicates a combination of a small blink, or saccade, with a lateral eye movement to the left.   This is also reflected in the voltage map. If it were a pure lateral movement, the inversion line would be vertical. Here we see it is more of a diagonal across the eye regions: 

                                                              06 EyeM Left Map

Image 6: Voltage map with an average reference applied to the data. Red is positive, blue is negative and white is neutral. Saturation of color indicates intensity, i.e. the darker the red, the more positive the value. Map projected onto 3D head where the data (color) in between sensors is interpolated. 

The more electrodes we have, the more detail we can glean about what exactly happened at any given moment. This subject blinked while they looked to the left! Similarly, soon after this event, we see a lateral eye movement to the right: 07 EyeM Right Chart

Image 7: bipolar montage of 10-20 equivalent electrodes; a subset from a 256 dense array EEG data set. Polarity is Negative up. Eye movement seen where the yellow line (Time Sync Marker) appears over the data. 

Since it is not also combined with an eye blink, this eye movement is not as obvious in the chart view. In the plot view below of the same time point, we see clearly the positive gradient on the right check electrodes, the direction the eye is moving towards, along with a negative gradient on the left cheek electrodes. Notice there is only a very small amount of positivity in the frontopolar electrodes, again indicating this is more of a pure lateral eye movement without the component of a blink. 

08 EM Right Plot

Image 8: High density EEG Topo Plot showing an eye movement to the right.

 

And the voltage map validates this as well, since the inversion line is closer to vertical:  

                                                           09 EyeM Right Map

Image 9: Voltage map with an average reference applied to the data. Red is positive, blue is negative and white is neutral. Saturation of color indicates intensity, i.e. the darker the red, the more positive the value. Map projected onto 3D head where the data (color) in between sensors is interpolated. 

 

If we can gather this much detailed information about what was going on during an artifact event, think of the amount of information we can gather about brain activity using high density EEG!


Additional Info

  • Product Type: Net Station 5
  • Information Type: Theoretical Background
Last modified on Monday, 24 April 2017 18:48

Return to Knowledge Center Home