Keywords
evoked potentials - auditory - speech perception - electrophysiology
Introduction
Cortical Auditory Evoked Potentials (CAEPs) have gradually entered clinical practice,
being useful to support diagnoses of central auditory disorders. Furthermore, this
assessment can reflect neuroeletrical activity of the auditory pathways.
CAEPs can be elicited by both verbal and non-verbal stimuli,[1]
[2] which reflect the neuroelectrical activity of the auditory pathway in the regions
of the thalamus and auditory cortex.[3] Several studies have aimed at analyzing electrophysiological assessment with speech
stimuli, including the verification of peripheral auditory structures, as in the Auditory
Brainstem Response (ABR).[4]
[5]
[6]
Some researchers suggest that verbal stimuli are ideal for studying the neural basis
of speech detection and discrimination.[1]
[2] Additionally, these types of stimuli can contribute to the assessment of complex
signals in the auditory cortex. Recent studies support the use of complex signals
for the evaluation of retrocochlear pathologies, central auditory disorders, and verification
of hearing aids.[7]
[8]
McPherson [9] and McGee [10] published normative values for the CAEPs with tone burst stimuli. Other studies
have also reported normative values for the tonal stimuli.[11]
[12]
[13] On the other hand, literature shows different values for verbal stimuli.[14]
[15] Authors describe the latency and amplitude of P300 as sensitive to the task demand,
and higher latency and lower amplitude for speech stimuli.[16]
[17]
Although the literature describes differences between tone burst and speech stimuli,
the protocols for verbal stimuli vary in their application, the stimuli used, and
the values of latency and amplitude. Thus, further studies should establish rules
and criteria so that this procedure can be effectively applied in clinical practice.
The aim of this study is to compare and describe the values of latency and amplitude
of CAEPs for speech stimuli in normal hearing adults, in order that the results may
serve as a reference for clinical and research in audiology and other areas.
Methods
The Research Ethics Committee of the University where this study was conducted granted
its approval for the study.
Participants who agreed to the research signed the term of responsibility. They received
information on all procedures from this research.
All participants were aged between 18 and 32, male and female, with normal hearing,
free from ear's disease or auditory process disorders, and without continuous use
of medication. They needed to be able to understand all procedures.
The authors excluded from this study individuals with hearing loss and auditory processing
disorder, guiding them to specific assessments.
All subjects underwent evaluation through anamnesis, visual inspection of the external
acoustic meatus, audiometry, middle ear assessment, and long latency auditory evoked
potentials with different verbal stimuli.
The anamnesis provided information on patientś audiological evolution and auditory
processing disorder.
The patients were subject to a visual inspection of the external auditory meatus using
a KlinicWelch-Allyn clinical otoscope (KlinicWelch-Allyn, NY, USA) to discard any
pathological changes that might have influenced audiometric thresholds.
The authors performed audiometry in an acoustically treated place using the audiometer
Itera II Madsen (Otometrics, Denmark). They tested the frequencies 250, 500, 1000,
2000, 3000, 4000, 6000, and 8000 Hz, using the descending-ascending technique. The
study considered as normal hearing individuals those with three-tone average (500,
1000, and 2000 Hz) less than or equal to 25 dB HL (decibel hearing level).[18]
The acoustic impedance measurements were performed by AT235 Interacoustics (Middelfart,
Denmark). All participants were submitted to tympanometric curve and acoustic reflexes.
The authors analyzed reflexes in the frequency range from 500 to 4000 Hz, bilaterally
in the contralateral mode. The study included only individuals with type “A” tympanogram
presenting acoustic reflexes.[19]
The CAEPs were performed with the Intelligent Hearing Systems (IHS), SmartEP module.
This equipment contains two response channels. The skin on all subjects was cleaned
with an abrasive paste. The electrodes were placed in the following positions: A1
(left mastoid) and A2 (right mastoid), Cz (vertex), the ground electrode (Fpz) on
the forehead. The impedance value for all electrodes was less than or equal to 3 kohms.
The patient received instructions to pay attention to different stimuli (rare stimuli)
that appeared within a series of equal stimuli (frequent stimuli). The percentage
of rare stimuli presented was 20%, while for frequent stimuli was 80%.
The speech tokens stimuli used were the consonant-vowel /ba/ as frequent stimuli in
all sequences, compared with different rare stimuli, as /ga/, /da/, and /di/. Therefore,
a sequence of different deviant stimuli were tested (/ba/ x /ga/, /ba/ x /da/, /ba/
x /di/). All speech token stimuli were presented in both ears at an intensity of 75 dB
HL. In total, 300 stimuli were used (60 rare and 240 frequent) to obtain the CAEPs.
The assessment began with /ba/ x /ga/, followed by /ba/ x /di/ and /ba/ x /da/. Prior
to obtaining the results, all participants received training to listen to the verbal
stimuli to become familiar with them. The patients had to report to the evaluator
the number of rare stimuli. The evaluator compared the response with the number of
rare stimuli effectively presented by the equipment. For the answer to be considered
correct there was a margin of error adopted of up to five stimuli that differed from
the exact amount presented by the equipment.
The authors obtained latency values for CAEPs to identify the waves in the greater
range and deflection peaks. They did not replicate results as this could tire the
individual and jeopardize the outcome of the assessment. The amplitude was measured
only for the P300 component, calculated from the baseline to the peak of the component.
The authors described and analyzed the results statistically using the Post Hoc Bonferroni test. They compared latencies of P1, N1, P2, N2, and P300 and the amplitude of P300
between the speech sounds. [Table 1] describes the parameters used in this study.
Table 1
Parameters used in this research of CAEPs with speech stimuli
|
Equipment
|
Intelligent Hearing System
|
|
Module
|
SmartEP
|
|
Electrodes
|
A1, A2, Fpz and Cz
|
|
Impedance of electrodes
|
Less or equal to 3 kohms
|
|
Type of stimulation
|
Binaural
|
|
Number of stimulus
|
300 (80% frequent and 20% rare)
|
|
Channels
|
AB
|
|
Rate
|
0.8 pps
|
|
Time
|
2.0 milliseconds
|
|
Phase
|
Alternating pattern
|
|
Speech tokens
|
/ba/ (frequent) /ga/ (rare)
/ba/ (frequent) /di/ (rare)
/ba/ (frequent) /da/ (rare)
|
|
Presentation of stimulus
|
Oddball paradigm
|
|
Time of stimulus
|
50.000 μs
|
|
Rise and decay time
|
20%
|
|
Envelope's stimulus
|
Trapezoidal
|
|
Individual state
|
Alert
|
Abbreviations: Kohms, kiloohms; ms, milliseconds; pps, pulses per second; μs, microseconds.
Results
In total, the researchers evaluated 30 subjects, of which 15 (50%) were male and 15
(50%) female. Their average age was 23.3 (±3.5) years.
Although stimulation has been binaural, the two-channel equipment allowed responses
for the right and left ears. The results were statistically analyzed using the Bonferroni post hoc test and no statistically significant difference were found between ears, both for
latency and amplitude. Therefore, to facilitate the analysis of this study, the authors
grouped the results of the right and left ears.
[Table 2] shows the percentage of presence of CAEPs for different stimuli. The other results
were obtained from all CAEPs considered present.
Table 2
Percentage of presence of cortical auditory evoked potentials with different speech
stimulus
|
Speech tokens
|
|
/ba/ x /ga/
|
/ba/ x /da/
|
/ba/ x /di/
|
|
N
|
%
|
N
|
%
|
N
|
%
|
|
P1
|
26
|
86.7%
|
27
|
90%
|
25
|
83.3%
|
|
N1
|
30
|
100%
|
30
|
100%
|
30
|
100%
|
|
P2
|
30
|
100%
|
30
|
100%
|
30
|
100%
|
|
N2
|
23
|
76.7%
|
16
|
53.3%
|
14
|
46.7%
|
|
P300
|
26
|
86.7%
|
28
|
93.3%
|
25
|
83.3%
|
Abbreviations: %, percentage of presence; N, number of subjects.
[Table 3] describes the latency values of CAEPs for all speech tokens. There was no statistically
significant difference between latencies of P1, N1, and P2. However, the latency for
N2 was greater for /ba/ x /di/ stimuli, and this difference was statistically significant.
For the P300 component, there was statistically significant differences between speech
tokens, being higher for the /ba/ x /ga/ stimulus.
Table 3
Average and standard deviation of latencies and amplitude for different speech stimulus
|
Speech tokens
|
|
/ba/ x /ga/
|
/ba/ x /da/
|
/ba/ x /di/
|
|
Average (ms)
|
SD (ms)
|
Average (ms)
|
SD (ms)
|
Average (ms)
|
SD (ms)
|
* p-value
|
|
P1
|
62.4
|
9.5
|
60.1
|
7.55
|
66.35
|
17.9
|
0.393
|
|
N1
|
103.55
|
10.45
|
103.5
|
11.4
|
108.55
|
18.05
|
0.038
|
|
P2
|
175.05
|
18.45
|
175.6
|
22.45
|
184.9
|
25.15
|
0.026
|
|
N2
|
250.5
|
33.3
|
234.8
|
41.05
|
256.5
|
35.45
|
0.006
|
|
P300
|
342.05
|
45.35
|
302.45
|
46.9
|
327.05
|
61.3
|
0.005
|
Abbreviations: ms, millisecond; SD, standard deviation.
*Post Hoc Bonferroni test.
[Table 4] describes the P300 amplitude values for different speech stimuli. There was no statistically
significant difference between stimuli.
Table 4
Average and standard deviation of P300 amplitude
|
Speech tokens
|
|
/ba/ x /ga/
|
/ba/ x /da/
|
/ba/ x /di/
|
|
Average (uv)
|
SD (uv)
|
Average (uv)
|
SD (uv)
|
Average (uv)
|
SD (uv)
|
* p-value
|
|
P300
|
6.4
|
2.15
|
7.35
|
5.35
|
6.5
|
2.65
|
0.208
|
Abbreviations: SD, standard deviation; uV, microvolt.
*Post Hoc Bonferroni.
[Table 5] shows the maximum and minimum descriptive values for the all variables studied.
Table 5
Maximum and minimum values of the variables
|
Speech tokens
|
|
/ba/ x /ga/
|
/ba/ x /da/
|
/ba/ x /di/
|
|
Max (ms)
|
Min (ms)
|
Max (ms)
|
Min (ms)
|
Max (ms)
|
Min (ms)
|
|
P1
|
86
|
56
|
64
|
50
|
76
|
50
|
|
N1
|
122
|
84
|
132
|
90
|
134
|
82
|
|
P2
|
224
|
136
|
220
|
150
|
226
|
142
|
|
N2
|
286
|
166
|
288
|
178
|
302
|
180
|
|
P300
|
430
|
210
|
430
|
220
|
446
|
236
|
|
Ampl P300(uV)
|
12
|
3.00
|
24.59
|
3.04
|
23.4
|
3.01
|
Abbreviations: Max, maximum; Min, minimum; ms, millisecond; uV, microvolt.
Discussion
Electrophysiological studies using complex stimuli have been increasingly prominent
in national and international literature. In general, more complex speech stimuli
evoke greater latencies and lower amplitudes of CAEPs. In addition, natural speech
stimuli evoke lower latencies compared with synthetic stimuli generated by the equipment.[20]
In this study, the percentage of presence for N1-P2 was greater than other CAEPs.
However, P1 and N2 were mainly affected by the speech tokenś characteristics. This
can be explained by the fact that the N1-P2 complex is the most visible exogenous
potentials, which makes it less variable in relation to stimuli.[21]
[22] Our results are consistent with another study,[8] in which the percentage of presence was lowest for N2. Research studies[23] report that the cortical components are influenced by cognitive experiences of the
individual throughout his life. Thus, better individual experiences with hearing,
cognition, and music produce the best results in the CAEPs, meaning greater amplitudes
and lower latencies.
In this study, the authors correlate their values of latency and amplitude with other
similar studies considering the standard deviation. When values from other studies
fall within up to two standard deviations from those presented herein, they are considered
concordant.
In comparing the latencies of CAEPs for different stimuli, only the N2 component presents
a significant difference, being higher for /ba/ x /di/. This component (N2) suffers
maturational influences, mainly from 5 to 10 years, reducing latency and amplitude.[24] The results from this research agrees with another study,[25] in which the N2 component also suffered influence according to the type of stimuli
presented. The authors emphasize that N2 is associated with attention to rare stimuli,
and depends on the complexity of the stimuli; thus, the higher the complexity, the
higher the latency.
For the P300, there was also difference between stimuli, with greater latency found
in the /ba/ x /ga/ stimulus. Another study[25] reported similar results in in which the latency of P300 was higher for stimuli
of greater complexity. Researchers[26] also reported increased latencies in more difficult tasks. In this study, the authors
did not investigate the spectral complexity of stimuli; nonetheless, participants
informally reported that the /ba/ x /ga/ stimuli was the most difficult to identify.
This may justify the results for P300.
The P300 amplitude depends on the tasks performed by the individual. The amplitude
increases in accordance with attentional parameters and receives influence from cognitive
disorders.[27] In our study, the overall average amplitude of P300 ranged between 5.35 and 7.35uV
(microvolts) for all the different stimuli, and the overall average was 6.75uV. Considering
up to two standard deviations (mean of all stimuli DP), our results are in agreement
with another study[17] in which the authors found mean values of 6.61 uV for P300. There is a variation
in the literature of 1.7 to 20uV, and many authors do not use these values in interpreting
the results because of the wide range of values described.[28] In our study, we also found ample variation for the P300's amplitude in the values
between all the different speech tokens.
Regarding the descriptive measures, some researches[17] propose that the latency of P300 for verbal stimuli must be between 289.57ms and
408.33ms. In our study, various speech stimuli were used. Nevertheless, considering
the average values between the three stimuli (323.85ms), our results agree with that
proposed by the aforementioned authors.
We found no published papers reporting the amplitude and latency of CAEPs for specific
stimuli speech, such as /ba/ x /ga/, /ba/ x /da/, and /ba/ x /di/. Thus, the table
below ([Table 6]) suggests values, norms, and comparisons based on the average, considering up to
two standard deviations for latency and up to one standard deviation for the amplitude
of P300. We determined only one standard deviation for the amplitude due to the wide
variation of results. These results are useful for future studies that use the same
speech tokens.
Table 6
Range latency and amplitude of cortical potentials obtained in this study
|
Speech tokens
|
|
/ba/ x /ga/
|
/ba/ x /da/
|
/ba/ x /di/
|
|
P1 (ms) (2SD)
|
43.4 - 81.4
|
45 - 75.2
|
30.5 - 102.1
|
|
N1 (ms) (2SD)
|
82.6 - 124.4
|
80.7 - 126.3
|
72.4 - 144.6
|
|
P2 (ms) (2SD)
|
138.1 - 211.9
|
130.7 - 220.5
|
134.6 - 235.2
|
|
N2 (ms) (2SD)
|
183.9 - 317.1
|
152.7 - 316.9
|
185.6 - 327.4
|
|
P300 (ms) (2SD)
|
251.3 - 432.7
|
208.6 - 396.2
|
204.4 - 449.6
|
|
Amplitude of P300 (uV) (1SD)
|
4.25 - 8.55
|
2.0–12.7
|
3.85 - 9.15
|
Abbreviations: SD, standard deviation; uV, microvolt.
The description and comparison of these values is important for clinical use. The
audiologist must understand the changes in cortical potentials for different stimuli.
The stimuli selected may compromise the results of the evaluations due to their complexity.
Therefore, the use of speech tokens is recommended as long as it is possible to know
the different results in our clinical practice.
Conclusion
This study demonstrates that the protocol for speech stimuli described produces similar
results from different stimuli, albeit the latency of N2 was higher for /ba/ x /di/,
while the P300 amplitude was greater for /ga/ x /ba/. Moreover, the description of
amplitude and latency values for different speech stimuli provide useful material
for future studies.
