EEG recordings are usually stored on paper or disk. There is a face sheet which is
attached to the paper record, if one exists, or is separate, in the case of digital recordings. For all recordings, the face sheet has all of the basic identifying information:
Figure 3-7: Bipolar and referential montages.
Common montages used in routine EEG. A: Electrode positions.
B: Longitudinal bipolar (LB) montage. C: Transverse bipolar (TB) montage. D: Referential montage.
• Name • Age
• Identification number of the patient • Index number of the recording • Reason for the study
• Name of the technologist • Current medications
• Time of the last seizure, if appropriate
• Technical summary, including activation methods and artifacts
• Technicians observations, including regions of particular concern or interest. • Time and date of the recording
• Ordering clinician
• Sedative medication used
Digital recordings should be identified on the face sheet, with the index number of the disk and the format of storage.
The face sheet should be filled out completely before the patient leaves the lab. If the technician sees a finding of immediate clinical concern, the neurophysiologist should be called immediately.
Calibration
Two phases of calibration are used prior to each study - square-wave calibration and biological calibration.
Square-wave calibration
A square-wave pulse is delivered from a wave-form generator into each amplifier input. This pulse is 50 µV in amplitude and alternated on and off at 1 second intervals. The wave does not appear a precise square wave because of the effects of the preset default filters. Figure 3-8 shows a sample recording.
The low-frequency filter (LFF) transforms the plateau of the square wave into an exponential decay. The high-frequency filter (HFF) slightly rounds off the peak of the calibration. For educational purposes, try several HFF and LFF settings during the calibration test to see the effects of filter changes on the record. It is also instructive to change filter settings during recording of EEG activity at a time which would not interfere with clinical interpretation.
The time constant (TC) of the LFF can be measured from the square-wave calibration page. TC is equal to the time it takes for a potential to fall to 37% of peak value. To eye- ball the value, this is approximately one third of peak value.
It is difficult to estimate the HFF setting from the square wave calibration, however, neurophysiologists should have an idea of what the peak should look like. If the HFF is
set too low, there will be a slow roll-off on the peak of the calibration pulse. If the HFF is set too high, the wave will appear too peaked and may even show overshoot, as if there is too little pen damping.
The experienced neurophysiologist can tell if there is a problem with one or more amplifiers by looking at the recordings made from each channel. Abnormalities are obvious if present in one channel; abnormalities in multiple channels may be more subtle. Types of abnormalities can include:
• Peak too rounded • Peak overshoot
• Incorrect rate of decay
If the peak is too rounded, the high- frequency response of the system is impaired. This usually means that there is excessive pen damping or the high-frequency filter setting is
improperly set. If the peak
overshoots, pen damping is usually set too low, allowing for inertia to produce excessive pen excursion. Incorrect rate of decay can mean that a filter setting is wrong for one or more channels or that the
amplifier is failing. Amplifiers may fail to the point that they give no response, but more commonly, they give distorted responses prior to failing completely.
Biological calibration
Biological calibration, of Biocal, assesses the response of the amplifiers, filters, and the recording apparatus on a complex biological signal. Electrodes Fp1 and O2 are connected to all of the amplifier inputs. The recordings from all of the channels should be identical. Pen pressure and damping
Pen pressure and damping are issues which apply only to paper recordings, digital
recordings are immune to these causes of distortion. Mechanical writing instruments have two inherent limitations: inertia and friction. Even when the filters are set properly, the frequency response may be inaccurate because of these mechanical factors. The physical mass of the pen produces inertia that slows its response time to sudden changes of signal voltage. Inertia is partially compensated for by control mechanisms in the pen drive mechanism. Friction is also compensated by EEG machine electronics, but excessive pressure of the pen on paper results in a sluggish response. This is the main reason that
Figure 3-8: Square-wave calibration
Square wave calibration signal responses with different filter settings.
A: Stadard filter settings
B: Reduction in the frequency setting of the low- frequency filters.
C: Increase in the frequeny settings of the high- frequency filter.
neurophysiologists and technicians need a visual memory of what a calibration pulse looks like with proper filter settings.
The same inertia that inhibits pen movement also promotes excessive pen movement, overshoot. Overshoot is minimized by the pen control mechanism, termed damping. When damping is not sufficient, normal waveforms may look like spike discharges. These effects are minimized with proper setting of pen pressure and damping. The manuals provided with EEG machines give instructions on setting of damping and pen pressure.
Sensitivity
The recording sensitivity is initially set at 7 µV/mm and subsequently adjusted depending on the amplitude of the EEG activity. Movement artifact and other non-cerebral
transients may exceed maximal pen excursion, but electrocerebral activity may not. Important waveforms may be missed when the sensitivity is set too low.
For children, the sensitivity is often reduced to 10-15 µV/mm because EEG amplitude is high in both the awake and sleep states. The elderly often have low-voltage EEG activity, and increased sensitivity is required.
Studies performed for the determination of brain death are started at 7 µV/mm but the sensitivity is always increased to 2 µV/mm.
Duration
A routine EEG should include at least 20 minutes of relatively artifact-free record. Longer duration recordings are often helpful in neonates so that the transitions between states can be identified. The Guidelines recommends 30 minutes of recording for brain death studies.
Filters
The standard filter settings for routine EEG are: • LFF = 1 Hz
• HFF = 70 Hz
The LFF of 1 Hz corresponds to a TC of 0.16 sec. If the LFF is set higher than 1 Hz, there will be attenuation and distortion of some slow waves. Slow waves have an increased number of phases and are composed of faster frequencies. Technicians should be discouraged from turning up the LFF especially when there is an abundance of slow activity. If the HFF is set too low, fast activity is blunted, and spikes and sharp waves may be impossible to identify.
The 60-Hz filter should not be needed in most laboratories. Three ways to minimize 60- Hz activity are:
• Selection of equipment location • Grounding
• Shielding
Shielding of the room is desirable but usually not essential and cannot completely abolish artifact induced by strong electromagnetic fields. Studies in a special care unit usually require use of the 60-Hz filter. Sources of artifact include ventilators, intravenous infusion pumps, air beds, heating or cooling blankets, and monitoring equipment. Activation methods
The performance and interpretation of records obtained with activation methods are discussed in Chapter 4. Hyperventilation, photic stimulation, and sleep may activate epileptiform activity. After an initial period of recording in the relaxed, wakeful state, the patient is asked to hyperventilate for 3 minutes. If absence seizures are suspected, the patient is asked to hyperventilate for 5 minutes. Hyperventilation is not performed in elderly individuals or in patients with advanced atherosclerotic disease because of concern for vasoconstriction with resultant cardiac or cerebral hypoperfusion. Photic stimulation is performed on older children and adults of all ages. Photic stimulation of sleeping infants is probably of limited clinical value.
Sleep is not considered by some to be a true activation method, because it is a transition between natural states. However, sleep helps to promote epileptiform activity, and in routine EEG, sleep frequently has to be induced by sedatives. Sleep deprivation may be needed if sedated sleep is not obtained. In this sense, sleep is an activation method. Sleep recordings are routinely indicated for all patients being evaluated for seizures, but are not helpful for patients being evaluated for encephalopathy. The mechanism of sleep is not important. There is no convincing evidence that natural sleep, sedated sleep, and sleep deprivation differ in their ability to promote epileptiform activity.
Electrode impedance
Electrode impedance should be at least 100 ohms and no more than 5 kohm. Lower impedances cannot be obtained with scalp electrodes without there being some improper connection between the electrodes, such as smeared gel between adjacent electrodes. Higher electrode impedances can create impedance mismatch which can then degrade the rejection of artifact by bipolar recordings.
Electrical interference is common in EEG laboratories, but since most recordings are bipolar, even referential recordings, the noise is typically similar in conformation and amplitude in the leads. Since the bipolar recording setup subtracts the voltage of the reference from the active signal voltage, the noise cancels out. Signal which is in
common to the two inputs is called common mode. This rejection can be described by the common mode rejection ratio. This represents the ability of an amplifier to reject signal in common to two inputs of the the amplifier.