Following a transient acoustic stimulus, such as a click or a brief tone pip, the ear and parts of the nervous system generate a series of electrical signals with latencies ranging from milliseconds to hundreds of milliseconds. These auditory evoked potentials (AEPs) are conducted through the body tissues and can be recorded from electrodes placed on the skin to evaluate noninvasively the function of the ear and portions of the nervous system activated by the acoustic stimulation. These short-latency or brainstem auditory evoked potentials (BAEPs) have proven to be valuable tools for hearing assessment, diagnosis of neurologic disorders, and intraoperative monitoring.
Overview of auditory evoked potentials
AEPs have been divided into short-latency components, with latencies of under 10 msec in adults; long-latency AEPs, with latencies exceeding 50 msec; and middle-latency AEPs, with intermediate latencies ( Fig. 24-1 ). The earliest components derive from electrical processes within the inner ear and action potentials in the auditory nerve. AEP components generated within the brainstem may reflect both action potentials and postsynaptic potentials. Auditory-evoked neural activity becomes affected increasingly by temporal dispersion as the poststimulus latency increases and as the contribution of short-duration electrical phenomena (e.g., action potentials) is eliminated. Thus, AEP components that are longer in latency are also wider, and the middle- and long-latency AEP components are generated predominantly by postsynaptic potentials within areas of cerebral cortex that are activated by the acoustic stimulus. AEP components are also affected increasingly by the state of the subject and by anesthesia as their latency increases.
Long-latency AEPs are affected profoundly by the degree to which the subject is attending to the stimuli and analyzing stimulus features ( Fig. 24-2 ). They have therefore been used as probes of cognitive processes, but their variability, as well as uncertainty about the precise identity of their cortical generators, limits their utility for neurologic diagnosis. Middle-latency AEPs are small; are subject to contamination by myogenic signals; and are rather variable from subject to subject, which also limits their clinical application. Both middle- and long-latency AEPs are affected prominently by surgical anesthesia.
Short-latency AEPs have achieved the greatest clinical utility because they are relatively easy to record and their waveforms and latencies are highly consistent across normal subjects. They are unaffected by the subject’s degree of attention to the stimuli and are almost identical in the waking and sleeping states, aside from minor differences related to changes in body temperature. Sedation also produces only minor changes in BAEPs, and has been employed during BAEP recordings. However, the use of sedation for evoked potential recordings has been reduced markedly owing to concerns about monitoring and the care of patients during conscious sedation. Because a typical surgical level of anesthesia produces only minor alterations in BAEPs ( Fig. 24-3 ), they can be used for intraoperative monitoring of the ears and the auditory pathways.
The earliest electrical signals produced by the auditory system in response to a transient stimulus, constituting the electrocochleogram (ECochG), initially were recorded by electrodes placed directly in the middle ear. Extratympanic recordings of the ECochG yielded smaller signals that required signal averaging to achieve an adequate signal-to-noise ratio. Signal averaging also permitted recording of small time-locked signals originating from other locations. Additional deflections with latencies of several milliseconds after auditory stimulation were recorded first in humans during signal-averaged ECochG studies by Sohmer and Feinmesser. Jewett and co-workers identified the short-latency scalp-recorded AEPs as far-field potentials volume-conducted from the brainstem, described the components and their properties, and established the Roman numeral labeling of the peaks that is used in most laboratories ( Fig. 24-4 ). A far-field potential is a potential (voltage) recorded at a sufficiently large distance from its source that small movements of the recording electrode have no significant effect on the waveform.
Although short-latency AEPs commonly are called brainstem auditory evoked potentials, this term is not completely accurate because the roster of generators clearly includes the distal (with respect to the brainstem) cochlear nerve and may also include the thalamocortical auditory radiations, neither of which is within the brainstem. Nonetheless, the designation BAEP is used in this chapter because it is the most widely used and understood term. Other synonyms or related designations include auditory brainstem response, far-field electrocochleography, and brainstem audiometry.
Standard baep recording techniques
This section describes the standard techniques used to record BAEPs in adult subjects. BAEP recording techniques for infants and children are described in Chapter 25 , and intraoperative BAEP monitoring is discussed in Chapter 30 . The American Clinical Neurophysiology Society has published guidelines for clinical BAEP recordings.
BAEPs are elicited most commonly by brief acoustic click stimuli that are produced by delivering monophasic square pulses of 100-μsec duration to headphones or other electromechanical transducers at a rate of about 10 Hz. A rate of exactly 10 Hz or another submultiple of the power line frequency should be avoided; otherwise, the inevitable line frequency artifact will be time-locked to the stimuli and will not be removed by the averaging process. Audiometric headphones having a relatively flat frequency response are desirable so that “broad-band” clicks, whose energy is spread over a wide frequency range, will be produced.
The stimulus intensity should be loud enough to elicit a clear BAEP waveform without causing discomfort or ear damage; 60 to 65 dB HL is a typical level. If hearing loss is present, stimulus intensity may be adjusted accordingly so that stimulation is at 60 to 65 dB SL. (It should be noted that dB HL is decibels relative to the threshold of a normal population, dB SL is decibels relative to the threshold of the ear being tested, and dB nHL is decibels relative to the threshold of the specific control population used to establish a laboratory’s normative database.) Reduced stimulus intensities are also useful during BAEP recordings, as discussed later in this chapter. The subjective click threshold should be measured in all subjects in whom this is possible, to recognize hearing loss and to determine the stimulus level corresponding to 0 dB SL.
Stimuli are delivered monaurally, so that a normal BAEP to stimulation of one ear does not obscure the presence of an abnormal response to stimulation of the other ear. An acoustic stimulus delivered to one ear via headphones can reach the other ear via air and bone conduction with a volume attenuation of 40 to 70 dB and generate an evoked potential by stimulation of the contralateral ear ( Fig. 24-5 ). To prevent this contralateral stimulation from occurring and possibly being misinterpreted as a BAEP arising from stimulation of the ipsilateral ear, the contralateral ear is masked with continuous white noise at an intensity 30 to 40 dB below that of the BAEP stimulus. Acoustic crosstalk also occurs with ear-insert transducers, though the signal reaching the opposite ear is attenuated to an even greater extent, typically 70 to 100 dB.
If the electrical square pulse causes the diaphragm of the acoustic transducer to move toward the patient’s ear, a propagating wave of increased air pressure, termed a compression click (also called a condensation click ) is produced. Reversing the polarity of the electrical square pulse that activates the transducer produces a rarefaction click . BAEPs to rarefaction and compression clicks may differ (most prominently in patients with a cochlear high-frequency hearing loss), and averaging the responses to the two click polarities together may produce a composite waveform with less diagnostic utility. Therefore, a single stimulus polarity should be used unless alternating polarities are necessary for canceling the electrical stimulus artifact or cochlear microphonic. Rarefaction clicks are generally preferable because they tend to yield BAEPs with better definition of the components; this may be because the initial cochlear movements produced by rarefaction clicks are in a direction that depolarizes the hair cells. In evaluating a patient’s BAEPs, the normative data used should have been acquired with the same stimulus polarity used to test the patient.
Recording electrodes typically are placed at the vertex (location Cz of the International 10–20 System) and at both ear lobes. (The ear lobes ipsilateral and contralateral to the stimulated ear are labeled Ai and Ac, respectively.) Electrodes at the mastoids (labeled Mi and Mc) may be substituted for the ear-lobe placements, although wave I tends to be smaller. Such placement, in combination with increased pickup of muscle noise, yields a poorer signal-to-noise ratio for wave I with mastoid leads than with ear-lobe leads. The ground electrode is often placed on the forehead, but its precise location is not critical. Metal cup or pellet electrodes may be used; needle electrodes should be avoided. Electrode impedance should be less than 5 kohm. Optimally, the same type of electrode should be used at all recording positions, and electrode impedance should be as consistent as possible across all recording electrodes, because mismatched electrode impedances can increase the amount of noise in the BAEP data.
BAEPs should be recorded between Cz and either Ai or Mi. A minimum of a two recording-channel system, with Cz–Ac or Cz–Mc in the second channel, has been recommended because this channel may aid in the identification of waves IV and V, which may be fused in the channel 1 waveform. An Ac–Ai or Mc–Mi channel may assist in the identification of wave I, and can be substituted in channel 2 when necessary, or can be recorded as a standard part of all tests if more than two recording channels are available.
The raw analog data are amplified by high input-impedance differential amplifiers with a common mode rejection ratio of at least 80 dB (10,000 : 1). A typical analog filter bandpass is 100 Hz or 150 Hz to 3,000 Hz (−3 dB points). The analog gain depends on the input window of the analog-to-digital converter; a value of approximately 100,000 is typical.
Data typically are digitized over an epoch duration or analysis time of approximately 10 msec. (The analysis time in some recording systems is actually 10.24 msec.) However, a longer analysis time of 15 msec may be required for recording pathologically delayed BAEPs, BAEPs to lowered stimulus intensities (as when recording a latency–intensity study), BAEPs in children, and BAEPs during intraoperative monitoring. The Nyquist criterion requires analog-to-digital conversion with a sampling rate of at least 6,000 Hz (2 × 3,000 Hz) in each channel to avoid aliasing, but higher sampling rates are required for accurate reproduction of the BAEP waveform and precise measurement of peak latencies. The analog-to-digital conversion should use at least 256 points per epoch; sampling of a 10.24-msec epoch at 256 time points corresponds to a sampling interval of 0.4 msec and a sampling rate of 25,000 Hz.
Far-field BAEPs are too small to be visible in unaveraged raw data, and so signal averaging is required. The improvement in the signal-to-noise ratio is proportional to the square root of the number of data epochs included in the average. Automatic artifact rejection is used to exclude sweeps with high-amplitude noise from the average. The number of epochs per trial is typically 2,000, although a larger number may be required if the signal-to-noise ratio is poor (usually reflecting a low-amplitude BAEP or noisy raw data). At least two separate averages should be recorded and superimposed to assess reproducibility of the BAEP waveforms. Latency replication to within 1 percent of the sweep time and amplitude replication to within 15 percent of the peak-to-peak amplitude have been recommended as standards for adequate reproducibility.
Waveform Identification and Measurement
When recording far-field potentials such as the BAEPs, the two electrodes connected to the inputs of a differential amplifier cannot be considered active and reference electrodes. The voltage distributions of the BAEPs extend over most, if not all, of the head, and no cephalic electrode can be considered to be truly inactive for all components. For example, wave V in the Cz–Ai BAEP waveform largely reflects positivity at the vertex, whereas wave I is derived from a negativity around the stimulated ear that is picked up by the Ai electrode. Thus, both input electrodes must be specified when describing a BAEP waveform.
The Cz–Ai BAEP typically is displayed so that positivity at the vertex relative to the stimulated ear is displayed as an upward deflection, and the upward-pointing peaks are labeled with Roman numerals according to the convention established by Jewett and Williston (see Fig. 24-4 ; Fig. 24-6 ). The downward-pointing peaks are labeled with the suffix N according to the peak that they follow; for example, downward peak IN follows upward wave I. The downward deflection following wave V, which has also been labeled the slow negativity (SN), is typically wider than the positive components and the earlier negative peaks.
Recognition and identification of the components in BAEP waveforms obtained with standard recording techniques will now be described. Modifications in the recording paradigm that can be used to enhance specific BAEP components are discussed later in this chapter.
The BAEP waveform typically begins with an electrical stimulus artifact that is synchronous with stimulus production at the transducer. Wave I is the first major upgoing peak of the Cz–Ai BAEP. It appears as an upgoing peak of similar amplitude in the Ac–Ai waveform and is markedly attenuated or absent in the Cz–Ac waveform (see Fig. 24-6 ). The cochlear microphonic may be visible as a separate peak preceding wave I, especially if the stimulus artifact is small. Its scalp distribution is similar to that of wave I. They may be distinguished by reversing the stimulus polarity, which will reverse the polarity of the cochlear microphonic; wave I may show a latency shift, but will not reverse polarity.
A bifid wave I is occasionally present and represents contributions to wave I from different portions of the cochlea. The earlier of the two peaks, which reflects activation of the base of the cochlea, corresponds to the single wave I that is typically present in the Cz–Ai waveform. Reversal of stimulus polarity can be used to distinguish a bifid wave I from a cochlear microphonic followed by (a single) wave I.
In contrast to wave I, wave IN is present at substantial amplitude in the Cz–Ac channel (see Fig. 24-6 ). This downgoing deflection is usually the earliest BAEP component in that waveform.
Wave II is typically the first major upward deflection in the Cz–Ac waveform because wave I is markedly attenuated or absent there. When present, wave II is usually of similar amplitude in the Cz–Ai and Cz–Ac channels (see Fig. 24-6 ). However, wave II may be small and difficult to identify in some normal subjects.
A substantial wave III is usually present in both the Cz–Ai and Ac–Ai channels. Wave III in the Cz–Ac waveform is usually substantially smaller than that in the Cz–Ai waveform (see Fig. 24-6 ). This difference helps to distinguish it from wave II, whose amplitude is little changed by the change in the ear electrode. The peak latency of wave III decreases slightly, whereas that of wave II increases slightly when the inverting amplifier input is changed from Ai to Ac; thus, the Cz–Ai waveform tends to give better separation of waves II and III than does the Cz–Ac waveform. A bifid wave III occasionally is observed as a normal variant ( Fig. 24-7 ); the wave III latency in such waveforms can be scored as midway between the peak latencies of the two subcomponents. Rarely, wave III may be poorly formed or absent in a patient with a clear wave V and a normal I–V interpeak interval; this finding is best interpreted as a normal variant waveform.
Waves IV and V are often fused into a IV/V complex whose morphology varies from one subject to another, and may differ between the two ears in the same person ( Fig. 24-8 ). The IV/V complex is often the most prominent component in the BAEP waveform. It usually is followed by a large negative deflection that lasts several milliseconds and brings the waveform to a point below the prestimulus baseline. Occasionally, the most negative point in the waveform follows wave VI (see Fig. 24-7 ), which could lead to misinterpretation of wave VI as an abnormally delayed wave V. The identity of wave V can be clarified by changes in recording montage or click polarity (see Figs. 24-6 and 24-7 ) or by reducing the stimulus intensity and/or increasing the stimulus rate to attenuate wave VI relative to wave V. In distinguishing between a totally fused IV/V complex and a single wave IV or V, Epstein notes that the former has a “base” that is greater than 1.5 msec in duration, whereas the width of a single wave is less than 1.5 msec.
When waves IV and V overlap in the Cz–Ai waveform, the wave V latency measurement used for BAEP interpretation should be taken from the second subcomponent of the IV/V complex, even if this is not the highest peak (in contrast to the amplitude measurement used to calculate the IV/V:I amplitude ratio, which is taken from the highest point in the complex). Measurement of the peak latency of wave V may be inaccurate if V appears only as an inflection on the falling edge of wave IV (see Fig. 24-8 ) and it may be impossible if they are fused smoothly. Two approaches may be used in such cases. The first involves measurement of wave V latency in a Cz–Ac recording channel. The overlapping peaks are separated more clearly there because the latency of wave IV is typically earlier, and that of wave V is later, than in the Cz–Ai waveform (see Fig. 24-6 ). However, because of these latency shifts, wave V latency values measured in a Cz–Ac waveform should be compared with normative data in which the latency of wave V was also measured in a Cz–Ac recording. The other approach is to reduce the stimulus intensity to attenuate wave IV relative to wave V and permit accurate measurement of the peak latency of wave V. That latency value cannot be compared with normative data obtained at a higher stimulus intensity, but the I–V interpeak interval can be evaluated because it is affected minimally by changes in stimulus intensity.
It is important to distinguish wave V from wave IV. If wave V were delayed abnormally, but an earlier and larger wave IV (which dominated the IV/V complex) were mistaken for wave V, the BAEP abnormality might be missed. If the latency of an apparent wave V is abnormally short, efforts should be made to determine whether this peak is, in fact, a dominant wave IV. Lesions that affect wave V almost always also affect wave IV, but rarely wave IV may be unaltered ( Fig. 24-9 ).
Clinical interpretation of BAEPs is based predominantly on the latencies of waves I, III, and V. Once these peaks have been identified and their latencies measured, the I–III, III–V, and I–V interpeak intervals are calculated. The latency of wave I has also been labeled the peripheral transmission time (PTT) and the I–V interpeak interval has been called the central transmission time (CTT). Differences of component latencies and interpeak intervals between stimulation of the right and left ears are also calculated.
The amplitudes of wave I and the IV/V complex are measured, each with respect to the most negative point that follows it in the waveform (I to IN and IV/V to VN), and their ratio is calculated. An excessively small IV/V:I amplitude ratio can identify as abnormal some BAEP waveforms in which all component latencies and interpeak intervals are normal ( Fig. 24-10 ). It sometimes has been called the V:I amplitude ratio. However, most studies establishing the value of this quantity measured the amplitude from the highest peak of the IV/V complex to VN, not from the wave V peak to VN. Measuring the latter may give an abnormally small ratio in normal waveforms in which the peak of wave IV is much higher than that of wave V.
Patient Relaxation and Sedation
The amplifier bandpass used for BAEP recordings filters out all of the delta, theta, alpha, and beta bands of the electroencephalogram (EEG), and the biologically derived noise in the recordings is derived predominantly from muscle activity. Therefore, patient relaxation during the recording session is essential to obtain “clean” waveforms with a good signal-to-noise ratio. Patients usually are tested while lying comfortably so that their neck musculature is relaxed. Patients should be requested to let their mouth hang open if the muscles of mastication are tensed. They are encouraged to sleep during testing because this aids relaxation and will not alter the BAEPs.
If the patient cannot relax sufficiently, sedation can be induced with agents such as chloral hydrate, a short-acting barbiturate, or a benzodiazepine; these have little or no effect on BAEPs in the usual sedative doses.
Modifications to baep recording techniques
Reducing the Stimulus Artifact
The transient electrical current in the transducer that generates the acoustic stimulus produces a transient electromagnetic field, which in turn induces a voltage in the patient and in the recording leads. This voltage creates an electrical stimulus artifact that, if large and prolonged, may overlap with wave I and impair the identification and measurement of that component. Using shielded headphones and headphones with piezoelectric transducers instead of voice coil transducers can reduce this artifact. Transducers that are connected to an ear insert by flexible plastic tubing several centimeters in length may also be used to mitigate this problem. The BAEPs, including wave I, are delayed by the time required for the acoustic stimulus to propagate through the plastic tube, but the electrical stimulus artifact remains simultaneous with transducer activation; the increased temporal separation between them permits greater decay of the electrical artifact before wave I. The greater distance between the transducer and the earlobe or mastoid recording site also serves to reduce the amplitude of the electrical stimulus artifact. The delay introduced by the plastic tube must be considered when one interprets BAEP latencies; passage through the tube may also change the acoustic properties of the stimulus. Ideally, the normative data used for BAEP interpretation should have been acquired with the same techniques that are used to test the patients. When ear inserts are used, they may be covered with metal foil to serve as near-field recording electrodes for wave I, thereby yielding a larger wave I.
Responses to compression and rarefaction clicks may be averaged together to reduce the electrical stimulus artifact by cancellation. Most evoked potential recording systems have an option for alternating the click polarity during stimulus delivery. This option should be avoided if possible during diagnostic testing because the BAEPs elicited by the two click polarities may differ. Stimulus artifact is usually more of a problem in the operating room environment, and alternating click polarity is often necessary during intraoperative BAEP monitoring. In this setting, the BAEPs are being compared with signals recorded earlier in the same patient with the identical stimulus paradigm, so the admixture of responses to the two stimulus polarities is less of a problem.
Improving the Resolution of Specific Components
Wave I may be small and difficult to record in some patients, especially if hearing loss is present. Several techniques can be used to obtain a clearer wave I. Because wave I is a near-field potential, relatively small movements of the Ai/Mi recording electrode can have a substantial effect on its amplitude. Thus, alternative electrode positions around the stimulated ear can be used. An electrode within the external auditory canal yields an even larger wave I, and may take the form of a metal foil covering on an ear insert or a spring-leaf electrode that makes contact with the wall of the canal.
Although wave I appears predominantly because of the presence of a near-field negativity at the Ai electrode, the horizontal orientation of its dipole projects a small positivity to the contralateral ear. Accordingly, an Ac–Ai recording channel can yield a somewhat larger and clearer wave I than that in the standard Cz–Ai recording.
Modifications in stimulus parameters are also useful in obtaining a clearer wave I. Alternating stimulus polarity can be useful by attenuating a large stimulus artifact that is obscuring wave I or by helping to differentiate wave I from the cochlear microphonic. A reduction in the stimulus repetition rate may yield a clearer wave I. Increasing the stimulus intensity is particularly helpful in improving the clarity of wave I and may be used if that peak is not clearly identifiable at standard intensities in the presence of hearing loss. One caveat is that in patients with a high-frequency hearing loss of cochlear origin, this maneuver yields interpeak intervals that are shorter than those of normal subjects. This difference in interpeak intervals probably reflects contributions to wave I from different portions of the cochlea. At normal intensities, the earliest peak, which reflects activation of the base of the cochlea, predominates. At higher intensities, a somewhat longer-latency component becomes increasingly prominent. If the earlier component is missing because of high-frequency hearing loss, and if stimulus intensity is increased, the wave I that appears may predominantly reflect the longer-latency contribution; the I–III and I–V interpeak intervals are thus shorter than those that would have been measured from the usual wave I. Measurement of the PTT from this longer-latency subcomponent would decrease the sensitivity of the test for recognition of prolonged I–III or I–V interpeak intervals.
Wave V is often fused with wave IV into a IV/V complex in the Cz–Ai BAEP waveform. A Cz–Ac recording channel helps to identify wave V by increasing the separation between the peaks of these two components; the latency of wave IV is shorter and that of wave V is longer than in the Cz–Ai recording (see Fig. 24-6 ). Because of this latency shift, the latency of wave V in a patient’s Cz–Ac waveform should be compared with normal values derived from Cz–Ac recordings in control subjects.
Modifications of the stimulus parameters can also be used to identify wave V if it is unclear when standard recording techniques are used. Wave V is the BAEP component most resistant to the effects of decreasing stimulus intensity ( Fig. 24-11 ) or increasing stimulus rate. If either of these stimulus modifications is performed progressively until only one component remains, that peak can be identified as wave V and then traced back through the series of waveforms to identify wave V in the BAEP recorded with the standard stimulus. Occasionally, wave V may be present following stimulation with one click polarity but not the other ( Fig. 24-12 ). Therefore, recording a BAEP with the opposite stimulus polarity may be useful if wave V is not identifiable with the standard laboratory protocol.
Stimulation at Several Intensities
In a patient with a conductive hearing loss, the stimulus intensity reaching the cochlea is less than that delivered to the external ear, and an abnormal BAEP with a delayed or absent wave I may result. If the stimulus intensity is increased to compensate for the conductive hearing loss, and no coexisting sensorineural hearing loss is present, a normal BAEP will be recorded. In contrast, BAEPs that are delayed as a result of abnormally slowed neural conduction do not normalize when the stimulus intensity is increased ( Fig. 24-13 ). Thus, increasing the stimulus intensity can help to differentiate peripheral from neural abnormalities, especially when wave I is not clear. The degree of hearing loss can be estimated by the elevation of the subjective auditory click threshold; increasing the stimulus intensity to compensate for this elevation is thus equivalent to stimulating at a specific intensity above that threshold (e.g., at 60 to 65 dB SL rather than at 60 to 65 dB HL). The normative data to which such recordings are compared should also have been recorded at a specific intensity in dB SL, rather than dB HL.
If each ear is stimulated at several different intensities, and the latency of wave V is graphed as a function of stimulus intensity, a latency–intensity curve is produced (see Fig. 24-11 ). Examination of latency–intensity curves may help to classify a patient’s hearing loss. A shift of the curve to a higher intensity level without a change in its shape suggests conductive hearing loss, whereas a change in the shape of the curve with an increased slope suggests sensorineural hearing loss. Latency–intensity curves may also reveal abnormalities that are not demonstrated by BAEP recordings at the standard, relatively high stimulus intensity typically used in BAEP studies performed for neurologic diagnosis.
A stimulus rate of approximately 10 Hz is used for routine clinical testing; this is because some of the BAEP components become attenuated and less clearly defined and the interpeak intervals lengthen as the rate is increased substantially above this. Measurement of the BAEP threshold is based solely on the presence or absence of wave V because it is the last BAEP component to disappear as the stimulus intensity is reduced. Because wave V is relatively resistant to the effects of rapid stimulation, recordings made solely for threshold measurement can be accomplished more rapidly by increasing the stimulus rate to 50 to 70 Hz. However, rapid stimulation can make BAEPs undetectable, especially in premature infants. Thus, infants who appear to have a hearing loss on BAEP screening using rapid stimulation should be retested using a slower stimulation rate.
When stimuli are delivered at a rate of about 10 Hz, the neural elements generating the BAEPs have approximately 100 msec to recover after responding to one stimulus before they must respond to the next. It has been speculated that, in some pathologic states, action potential propagation or synaptic transmission occurs normally with recovery times of this magnitude but is delayed when recovery times are shorter. If this were the case, more rapid stimulation would demonstrate BAEP abnormalities in some patients with normal BAEPs to the standard stimuli, thus increasing the sensitivity of the test. Rapid stimulation has been reported to increase the test sensitivity of BAEPs in patients with neurologic abnormalities in some, but not all, studies.
Alternative Auditory Stimuli
The standard BAEP stimulus is a “broad-band” click produced when an electrical square pulse is delivered to the headphone or other electromechanical transducer. It is generated predominantly by the region of the cochlea responding to 2,000- to 4,000-Hz sounds, although wave V may also receive contributions from lower-frequency regions of the cochlea. BAEPs have also been recorded following stimulation with brief tone pips, in an effort to probe specific parts of the cochlea. This method can be used to determine thresholds at different frequencies for BAEP audiometry. Tone pips most often are generated with stimuli consisting of brief bursts of sine waves. The burst cannot be stopped and started abruptly because an audible “click” containing many frequencies would be produced. Instead, the sine wave stimulus is amplitude-modulated with a rise-time, plateau, and fall-time to reduce (although not eliminate) the energy content of the stimulus at other frequencies. A filtered click, which is obtained by passing a broad-band click through a bandpass filter, has also been used as a frequency-specific stimulus.
Another technique that can be used to obtain frequency-specific information from BAEPs is acoustic masking. In one approach, the broad-band clicks used to elicit the BAEPs are superimposed on white noise that has been highpass-filtered by using different cut-off frequencies in different runs. Each BAEP waveform is generated predominantly by the unmasked region of the cochlea and represents a response to frequencies lower than the cut-off frequency used during that run. Subtraction of one such response from another yields a “derived” response that estimates the response to the frequency band between the cut-off frequencies of the two runs ( Fig. 24-14 ). The subtraction process yields waveforms with relatively poor signal-to-noise ratios, however.
Frequency-specific stimuli and frequency-specific masking can be combined by embedding tone pips in continuous notched noise. The latter is white noise that has been notch-filtered to remove the basic frequency of the tone pip. Shaping a pure tone into a pip with an onset and an offset introduces power at other frequencies into its frequency spectrum, which would impair the frequency specificity of the test, but the notched noise masks the responses to frequencies other than the basic frequency of the tone pip. BAEP audiograms produced with such stimuli are similar to behavioral audiograms in the same subject. Wave V is broader in the frequency-specific BAEPs elicited by the lower frequencies; when these are recorded, reducing the low-cut (high-pass) analog filter in the evoked potential averager to 30 Hz may yield a clearer wave V and more reliable results.
Stimulation with relatively prolonged, low-frequency tone bursts (e.g., 200- to 500-Hz tone bursts lasting 10 to 15 msec each) produces an evoked potential that contains oscillatory components at the frequency of the tone ( Fig. 24-15 ). The frequency-following response and the BAEPs do not appear to have the same neural generators. Whereas the BAEPs reflect activity originating in the base of the cochlea, the frequency-following response predominantly reflects activity originating in lower-frequency regions of the cochlea. However, the clinical utility of frequency-following responses for assessing hearing in the lower frequencies has been limited by the technical difficulties associated with recording them and by their relatively high thresholds.
BAEPs may also be elicited by bone-conducted stimuli. This is most useful in assessing patients who may have conductive hearing losses, such as neonates in whom BAEPs performed with air-conducted stimuli are suggestive of a hearing loss.
BAEPs to Electrical Stimuli
BAEPs can also be recorded following electrical stimulation of eighth nerve fibers through the electrodes of a cochlear prosthesis. This can be used to assess the proximity of these electrodes to the spiral ganglion during implantation and the adequacy of eighth nerve stimulation during programming of the processor, though recording of the evoked eighth nerve compound action potential through a different set of electrodes may prove to be a more useful tool for the latter application. Recording of BAEPs to electrical stimulation at the promontorium has proven to be a poor predictor of the success of cochlear implantation in children. However, recording of BAEPs to electrical stimulation via an intracochlear electrode has been shown to correlate well with auditory outcomes, and may prove to be useful in guiding therapy in young children with questionable auditory nerve integrity.
Magnetic Recording of BAEPs
Since the BAEP generators are far from the surface of the head, it is difficult to record the magnetic fields corresponding to the BAEPs, and such recordings are not practical for clinical use. However, prolonged averaging with a large number of epochs per average can yield magnetic signals within the latency range of the BAEPs. Studies of these signals may produce evidence about the generators of the BAEPs. The most consistent peaks of the magnetic BAEPs correspond in latency to waves IN and V of the electrical BAEPs that are elicited by the same acoustic stimulus.
Effects of varying stimulus parameters
Responses to rarefaction and compression clicks delivered to the same ear may differ (see Fig. 24-7 ). Component latencies following rarefaction clicks are usually, although not always, shorter than those following compression clicks. The BAEP latency differences between the two stimulus polarities tend to be greatest in patients with cochlear high-frequency hearing loss. BAEP peaks tend to be clearer with rarefaction clicks because of fusion of adjacent components when compression clicks are used, although again there are exceptions. The presence of wave V following one click polarity, but not the opposite, has been reported as an uncommon finding in patients referred for BAEP studies (see Fig. 24-12 ); the significance of this finding is unclear.
As the stimulus intensity is decreased, the latencies of all BAEP components increase (see Fig. 24-11 ). The increase is caused predominantly by a latency shift of wave I; the interpeak intervals are relatively constant until the stimulus intensity nears threshold. Component amplitudes decrease as the stimulus intensity is lowered, but with different patterns. Wave I shows the most rapid attenuation as the stimulus intensity is lowered, and it usually is lost before waves III and V. Wave V is attenuated to a lesser degree at lower stimulus intensities and is the last peak to disappear; its disappearance defines the BAEP threshold. At lower intensities, the IV/V:I amplitude ratio is increased because of the greater degree of attenuation of wave I.
Increasing the stimulus rate increases both the absolute latencies of the BAEP components and the interpeak intervals. As the stimulus rate is increased above approximately 10 per sec, component amplitudes decrease and the peaks tend to become less well defined. Wave V is most resistant to these effects. Thus, a stimulation rate of approximately 10 Hz is preferable for routine clinical testing, but more rapid rates may be used to facilitate recordings to measure the wave V threshold.
Generators and scalp topographies of BAEPs
In their first detailed description of BAEPs, Jewett and Williston hypothesized that most of the BAEP components represented composites of contributions of multiple generators. This premise is supported by the bifid wave III seen in some normal BAEPs (see Fig. 24-7 ), which contains two subcomponents that overlap in most subjects. The effects of variations in stimulus parameters on the topography of wave II suggest that this component arises from more than one generator. Shifts in the latencies of many BAEP peaks across various scalp recording locations (see Fig. 24-6 ) indicate that these peaks represent summations of overlapping components. Intraoperative recordings in humans and detailed intracranial mapping studies in animals have also demonstrated multiple generators for many BAEP components.
The complexity of the generators of human BAEPs ( Fig. 24-16 ) derives in part from the pattern of connections within the auditory pathways, with ascending fibers both synapsing in and bypassing various relay nuclei. It also reflects the presence of two bursts of activity in the auditory nerve, seen as the N1 and N2 action potentials of the ECochG, which can drive the more rostral pathways. Because of both of these factors, several different structures within the infratentorial auditory pathways may be active and may generate field potentials simultaneously.