Advantages of Digital Electroencephalography in Clinical Veterinary Medicine-1
Terrell A. Holliday and Colette Williams
Veterinary Medical Teaching Hospital, School of Veterinary Medicine, University of California Davis, Davis, California
This is the first of a series on digital electroencephalography. The present discussion deals with:
1. Changing electrode montages for display after recording is completed (so-called "remontaging"). (To read about the principles of forming electrode derivations and montages visit Holliday and Williams, VNN, 1999.)
2. Changing frequency limits after recording is completed.
1. Changing Electrode Montages After Recording ("remontaging").
Veterinary EEG recording typically is performed under certain limitations. Because of the cost of equipment, most veterinary laboratories have had polygraphs with only 6 or 8 EEG channels. By comparison, thirty-two channels are now used in routine recordings from human patients and even more in some specialized studies. In addition to monetary limits, there is an inherent limit on the number of electrodes that can be placed on the head of most veterinary patients, whose calvaria are often smaller even than those of children.
If a larger number of EEG channels is available, it is possible to use more electrodes to form more derivations and thereby make possible more precise localization of EEG events. When this is impossible, it is very helpful to use multiple montages that include the desired additional derivations. We formerly used as many as 5 montages to help deal with this problem. (Holliday and Williams, VNN, 1999)
Unfortunately, using multiple montages requires additional recording time and the animal's cooperation may not be sustainable for the necessary time. Sedation can be used and has distinct advantages, (Holliday and Williams, VNN, 1999) but sedative drugs have limited durations of action that also can limit the recording period. Digital EEG eliminates these problems. It allows one to record a single period of EEG and then, using the EEG data stored in digital format in the machine, "remontage" at will at any later time, thus effectively extending the recording period without limit. Any number of standard montages can be used and one can improvise extemporaneous montages if necessary to better interpret the data in individual cases.
Remontaging consists of changing the montage of an EEG after recording is completed. This is demonstrated in Figures 1-4. These figures all show exactly the same epoch of EEG. They differ from one another only in the ways the EEG has been remontaged. Recorded from a 4 week-old foal, the epoch includes a paroxysmal discharge (PD) (specifically, repetitive spike-and-wave complexes). For more information on EEG technics see Electroencephalographic Technics at UCD-VMTH Electrophysiology Laboratory section below.
Electrophysiology Notes - Advantages of Digital Electroencephalography-1
Click on the image to see a larger view
Figure 1 displays the PD with a commonly used 8 channel montage (Redding and Knecht, 1984). based on 5 electrodes. Although the PD is conspicuous, the EEG does not clearly indicate whether it is focal, widespread or even possibly generalized.
This EEG was recorded from a 4 week old foal. In the center of the figure, an approximately 3 sec-duration paroxysmal discharge (PD) consisting of spike-and-wave complexes, is visible in all derivations; in some derivations the complexes are "downpointing".
The EEG is displayed in a widely-used montage in which 5 electrodes are connected to the amplifiers to form 8 derivations (Redding and Knecht, 1984) . The first 4 derivations are "common reference derivations" (i.e., all using the same reference, in this case, the vertex electrode) the last 4 are "bipolar derivations" (See Electroencephalography Electrode Map section below). (For discussion of montages and derivations, see Holliday and Williams, VNN, 1999). In this montage it appears the PD probably occurred on the right side, rostral to the right occipital electrode and caudal to the right frontal electrode. The absence of electrodes between the frontal and occipital electrodes and the absence of electrodes in more lateral sites makes it impossible to localize more precisely. (for information on "locating" spikes and waves see Holliday and Williams, 1999) Eye movements accompanied this PD, but did not regularly occur with other PD in the record. Here only the right lateral canthus (OD) electrode was functional, the left one having come loose during recording, causing excessive artifact)
The foal's behavior during this period, and the EEG background rhythms (BGR) at the time, were those of REM sleep. The BGR show the high amplitudes expected in young mammals (note display amplification, shown in calibration bar, is less than the "customary" 50 microvolts).
In Figure 2, eight electrodes were used in 3 chains extending rostrocaudally, once again displayed in 8 channels. This montage makes it possible to better localize events by forming the electrode derivations into "chains" which take advantage of "instrumental phase reversal" to help in localizing focal events (Holliday and Williams, VNN, 1999). In this montage, it is apparent the event arose focally in the right hemisphere, with highest amplitude at or near the C4 electrode. A transverse chain of electrodes might allow even more precise localization
The same epoch shown in Figure 1 is shown here after remontaging. Here the PD appears to consist of a ~3 sec burst of 6/sec spike-and-wave complexes that is most clearly defined in the C4-P4 derivation, but is obvious in others (some are downpointing).
Here the EEG is displayed in an 11-electrode montage, still using only 8 channels. The electrodes are arranged in 3 rostrocaudal chains: one 4-electrode chain on the left side, a 3 electrode chain on the midline, and a 4-electrode chain on the right side. (See Electroencephalography Electrode Map section below). The amplitudes suggest the PD occurred on the right side. A phase reversal at C4 suggests the focus is near C4 (compare channels 6 and 7). The use of additional electrodes has improved localization somewhat. However, derivations forming transverse chains across the hemispheres could be helpful in clarifying the laterality of the event. (See Figure 3)
In Figure 3, the montage used in Figure 2 has been modified by adding a chain of 3 electrodes that extends from the left side, across the midline and onto the right side, at right angles to the midline and passing through the vertex (Cz). The occipital electrodes have been omitted so that an 8 channel display is still possible. The extent and focal nature of the event are more strongly emphasized; however, it could have occurred some unknown distance lateral to the C4 electrode.
This figure shows the same epoch as that in figures 1 and 2. Remontaging has omitted the occipital electrodes and added a short transverse chain (C3-Cz; Cz-C4, channels 7 and 8)). There is phase reversal of the spike-and-wave complexes at the C4 electrode in the rostrocaudal chain (compare F4-C4 and C4-P4, channels 5 and 6). In the transverse chain, the spikes and waves are downpointing at C3-Cz (channel 7) but at Cz-C4, (channel 8) directionality is equivocal, with phase reversal of spikes but uncertain direction of the waves. An electrode lateral to C4 possibly could be helpful in interpretation. (See Electroencephalography Electrode Map section below).
In Figure 4, a more complete array of electrodes is displayed. This montage uses all the electrodes used in the previous montages plus electrodes in the temporal regions (A1, A2). It comprises 3 rostrocaudal longitudinal chains and 4 left-to-right transverse chains, the latter in rostral, central, parietal and occipital locations. Even more precise localization is possible (Figure 4).
The same epoch is shown here in a montage containing 17 EEG channels formed by 13 electrodes connected in three rostrocaudal chains: left (channels 1-3);, midline (channels 4,5) and right (channels 6-8) and 4 transverse chains: frontal (channels 9, 10); central, extending from the left temporal region to the right temporal region (channels 11-14) ; parietal (channels 14, 15); occipital (only 2 electrodes, channel 17). (See Electroencephalography Electrode Map section below).
Amplitude differences and phase reversals at the C4 electrode in the rostrocaudal chain on the right side indicate the PD is affecting the C4 electrode more than others in that chain. In the transverse central chain the downpointing spikes and waves at C3-Cz indicate Cz is affected more than C3. The "up-pointing" complexes in the C4-A2 derivation (channel 14) indicate the event affected the C4 electrode more than A2. At Cz-C4, as in Fig 3, the PD is still of equivocal direction. The event probably occurred where it affected both electrodes about equally. Importantly, though it is visible in all channels, the PD does not appear to be generalized.
Electroencephalography Electrode Map
The electrode "map" shown here is that presently used for placing electroencephalography electrodes on horses at the UCD-VMTH Clinical Neurophysiology Laboratory.
Electrode designations conform, where possible, with internationally recommended terminology:
C: "central" **
Odd numbers on the left side,
Even numbers on the right side;
z: midline location.
Note: The A electrodes deviate from standard terminology in that they are located in the low temporal region, rostral to the ear canal and just above the level of the zygomatic arch, rather than being on the ear itself.
OD, OS, IC are electro-oculogram electrodes: OD, right eye; OS, left eye; IC, intercanthal, on the midline, between the medial canthi.
In the patient described herewith, the Fp electrodes were omitted because of the rather small size of the foal's head.
** We use the "C" designation because it is used widely in veterinary medicine even though it is inappropriate in all but primates. The C electrodes in higher primates are near the central sulcus of the cerebral hemispheres, hence their name. The analogous structure in the domestic species, the cruciate sulcus, is much smaller, much more rostral and is actually much closer to the frontal electrodes than to the C electrodes.
COMMENTS: Digital EEG permits greatly improved veterinary EEG results. Nevertheless, there is room for further improvement, for example: Present-day electrode montages, even with digital EEG, are limited now by two important factors: 1: space on the animal's head for more electrodes; 2. absence of recording from the more ventral regions of the hemispheres. We now have added additional electrodes in the lateral regions of the calvaria (not shown here) that seem to be helpful in more precise localization of abnormalities. Still, nasopharyngeal and/or sphenoidal electrodes are badly needed to allow recording from the lower and more rostral regions of the temporal lobes and the amygdala. We cannot truly "recognize" temporal lobe epilepsy in domestic animals and differentiate it from other epilepsies, without knowing the EEG status of such regions, interictally and ictally.
2. Changing Frequency Sensitivity to Reduce Muscle Artifact.
In domestic animals, most of the calvaria is covered by facial musculature. In EEGs recorded from such areas, high-frequency muscle artifacts sometimes abound and obscure the EEG. Sedation helps reduce the artifacts by allowing the animal to relax, but even in deep sleep, some muscle artifact can persist. Muscle potentials in the EEG tend to fall in the range of about 20 Hz to above 70 Hz. Most conventional ("hard copy") EEGs, use a high frequency limit of 70-90 Hz, therefore muscle potentials are readily recorded. Lowering the high frequency limit ("filtering") can reduce the muscle artifacts; however, when this is done the "hard copy" recording is permanently changed and the changes cannot be undone.
The high frequency limit on most digital EEG equipment is about 90 Hz, thus substantial amounts of muscle potentials will be included in the recording. Nevertheless, one can display the EEG later while applying any filter limits deemed necessary to reduce the muscle artifact. This simply affects the display of the EEG; the original data is maintained unchanged on the disk.
Changing high-frequency limits after recording.
Click on any image to see a larger view
Figure 5A shows an epoch of EEG displayed with a high frequency limit of 70 Hz. The recording is heavily laden with muscle artifacts, consisting of some low amplitude, steadily ongoing muscle potentials plus bursts of high amplitude, high frequency muscle potentials that occurred when the horse moved its mandibles and/or its ears.(Figure 5A)
This EEG epoch from a horse is heavily laden with ongoing high-frequency activity consisting of muscle potentials, e.g., channel 4. Superimposed on this in most channels are bursts of much higher amplitude high-frequency muscle potentials which occurred in association with ear movements and/or movements of the mandibles. Some low amplitude delta activity at about 1/sec is evident but it seems possible other slow activity is obscured by the artifacts.
The identical epoch is shown in Figure 5B except the high frequency limit has been reduced to 15 Hz. The low amplitude muscle potentials have been removed or greatly attenuated and even the high amplitude bursts are greatly reduced; this allows one to better appreciate the ongoing delta activity in the background rhythms (Figure 5B) (Readers please note: we do not advocate using such poor data in diagnosis, with or without lowering the high frequency sensitivity; we merely selected this as an extreme example.)
Figure 5B. The same epoch shown in Figure 5A is shown here after changing the high-frequency limit to 15 Hz. The artifacts have been nearly eliminated, allowing better visualization of the low frequency content of the recording. Note the delta and theta activity that was obscured or invisible before changing the frequency sensitivity. (See Electroencephalography Electrode Map section above).
"Filtering out the muscle" may appear to be "the best of all worlds", but to avoid errors of interpretation it is essential to recognize the changes filtering can induce: the "improvements" made by changing the high frequency limit are not without cost. In Figure 6, the upper trace (A) is displayed with the high frequency limit set at 70 Hz. In the lower trace (B) the high frequency limit is set at 15 Hz. Note that the spikes of the spike-and-wave complexes are obliterated or greatly attenuated and "blunted" by the filtration. This could result in failure to recognize spikes and misidentification of spike-and-wave complexes, especially if they occurred as isolated events rather than in conspicuous trains such as those in the figure (Figure 6). Take care in using filters!
A spike-and-wave complex displayed at 70 Hz high-frequency limit in the upper trace and at 15 Hz in the lower trace. Note that small spikes, such as those at A in upper trace, are eliminated or greatly attenuated at 15 Hz in lower trace (B). Larger spikes in the last 2 sec of the event are still present at 15 Hz but are smaller and not as sharp as when displayed at 70 Hz. This could lead to errors of interpretation if the frequency limits are not considered.
Electroencephalographic Technics at UCD-VMTH Electrophysiology Laboratory
In the UCD-VMTH laboratory we usually record 45-75 min of digital EEG. Recording is performed using 13-15 EEG electrodes; an additional 3 electrodes are used for the electro-oculogram (EOG) but can be used for EEG if desired. The ECG routinely is monitored from leads on the limbs (small animals) or the thorax (horses). Sedation is administered before placing electrodes. The foal whose EEG is shown herein was sedated with xylazine xxxmg/kg, IV.
Electrodes are platinum alloy needle electrodes placed subcutaneously in a routine manner, attempting to record from the dorsal and lateral surfaces of the cerebral hemispheres. For horses and medium-size or large dogs, 13 to 15 electrodes are placed and 13-19 EEG channels are recorded, plus the ECG and EOG (see Electroencephalography Electrode Mapsection above). The EEG equipment is programmed to record the EEG using the Cz (vertex) electrode** as a reference during recording. However, during recording and during interpretation of the record, the EEG can be displayed using any desired montage. During interpretation, montages are designed to take advantage of instrumental phase reversals (see Holliday and Williams, VNN, 1999) although from time to time referential montages are used.
Electrodes are connected to a Nihon-Kohden 2100 EEG system. Amplifier frequency range is 0.001 Hz to 90 Hz. For digitization, EEG signals are sampled at 200 Hz. All EEG in figures 1-4 are displayed with a time constant of 0.1 and a high frequency limit of 70 Hz, rolloff of 6 dB/octave and 12 dB/octave respectively. The high frequency limits in Fig 5 and Fig 6 were varied for demonstration purposes.
** We use the "C" designation because it is widely used in veterinary medicine even though it is inappropriate in all but primates. The C electrodes in higher primates are near the central sulcus of the cerebral hemispheres, hence their name. The analogous structure in the domestic species, the cruciate sulcus, is much more rostral and is actually much closer to the frontal electrodes than to the C electrodes.
1. Holliday, TA and Williams, DC. Interictal paroxysmal discharges in the electroencephalograms of epileptic dogs. Current Techniques in Small Animal Practice 13(3): 132-143, 1998.
2. Redding, RW and Knecht, CE. Atlas of Electroencephalography in the Dog and Cat. New York, Prager, 1984.