CHAPTER I INTRODUCTION 1.1 The problems Auditory Brainstem Response (ABR) (Hall, 1992) is one of the auditory evoked potentials obtained from the brain electrical activity through stimulation by an acoustic stimulation. It represents the neural activity from several anatomical structures within the peripheral and central auditory nervous system. ABR is composed of several voltage deflections occurring within the first 15 ms after stimulus onset (Katz, 2002) and consist of 5 to 7 peaks or waves that labeled using Roman numerals (Hood, 1998). ABR test offer several advantages such as assisting in identifying neurological abnormalities in the eight cranial nerve and auditory pathways of brainstem. Besides, ABR is
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
CHAPTER I
INTRODUCTION
1.1 The problems
Auditory Brainstem Response (ABR) (Hall, 1992) is one of the auditory evoked
potentials obtained from the brain electrical activity through stimulation by an acoustic
stimulation. It represents the neural activity from several anatomical structures within the
peripheral and central auditory nervous system. ABR is composed of several voltage
deflections occurring within the first 15 ms after stimulus onset (Katz, 2002) and consist
of 5 to 7 peaks or waves that labeled using Roman numerals (Hood, 1998).
ABR test offer several advantages such as assisting in identifying neurological
abnormalities in the eight cranial nerve and auditory pathways of brainstem. Besides,
ABR is also useful to estimate the hearing sensitivity of patients who cannot give valid or
reliable hearing threshold using behavioral methods (Hood, 1998).
Despite its advantages, the ABR has one obvious limitation. The ABR is a time
consuming procedure because several number of ABR signals need to be averaged and
collected in order to eliminate noise from other related activities. The process is called
signal averaging function to improve the ABR signal to noise ratio (SNR) thus improve
the likehood of ABR detection.
One of the possible solutions to reduce ABR acquisition time is by using high
stimulus repetition rate. High stimulus rate cause time between presentations of stimuli
equals to limit of ABR conventional averaging responses about 20 ms. However at rates
faster than limit, conventional averaging ABR response will overlap and interfere to one
another and causing distorted and useless final averaged (Eysholdt & Schreiner, 1982).
Hence, Maximum Length Sequences (MLS) paradigm was introduced in order to
increase the stimulus repetition rate that cannot be achieved using conventional
averaging. MLS is a pseudorandom binary sequence that allow for each subsequent
stimulus to be presented before the response of previous stimulus has completed. It works
by process of deconvolution where response recorded using MLS stimulation can be
extracted from the multiply overlapped combination of responses that will be generated
by the stimulus sequence.
Previous studies showed that high stimulus repetition rate through MLS generally
can improve the ABR testing time compared to the conventional stimulus rate. However,
by increasing the stimulus rate neural fatigue may cause poorer signal to noise ratio
(SNR). High stimulus repetition rate with poorer SNR may lead to difficulty for
Audiologist to detect ABR waveform with poor morphology. Thus, presenting the
stimulus very quickly does not turn out to be that efficient as the clinician need to present
more stimulus to get better SNR (Lasky R.E., Shi Y., & Hecox K.E., 1992; Marsh R.R.,
1992; Bell S.L., Allen R., & Lutman M.E., 2000).
However most of the literatures used an ideal SNR as a baseline instead of
minimal SNR to conclude that the high stimulus repetition rates through MLS paradigm
is not a beneficial tool to improve ABR testing time. In reality, Audiologist detects the
1
ABR waves based on its visibility rather than looking at the SNR value. Therefore there
is a possibility that MLS can improve the ABR testing time by looking to other
perspective such as Audiologist detection instead of looking at SNR.
1.2 Contributions of the study to the body of knowledge
This study contributed to the body of knowledge in audiology field specifically in
auditory electrophysiology part involving ABR test using MLS technique. Currently,
there is no study was conducted concerning from linear and non linear MLS with
subjective detection.
This study is the first report on the following areas:
1. In measuring the improvement of MLS testing time in infants.
2. Improvement in testing time provided by new non linear algorithm for infant.
2
CHAPTER II
LITERATURE REVIEW
2.1 Description of Auditory Brainstem Response
Auditory Brainstem Response (ABR) is an evoked responses occur within the
first 15 ms after stimulus acoustical or electrical stimulation consist of a series of 5 to 7
peaks or waves that represent neural activity at several anatomical sites (Hood,1998). The
ABR response is recorded by the placement of the electrode that is attached on specific
part of the head. ABR wave’s series are labeled based on the Roman numeral from I till
VII. In general, wave I represent the neural activity of distal or peripheral portion of
auditory nerve (Hood, 1998; Hall, 1992).Wave II is generated by the neural activity from
the proximal eight nerves as it enters the brainstem (Hood, 1998). However due to the
factor of age, wave II maybe absent in recording children as shorter eight nerve length
(Hall, 1992). Meanwhile, wave III is contributed by the neurons in the cochlear nucleus
and possibly other fibers that entering the cochlear nucleus (Hood, 1998). For wave IV,
studies suggest that the third-order neurons likely involved the superior olivary complex.
Other contributions of wave IV are the fibers at the area of cochlear nucleus and the
nucleus of the lateral lemniscus (Hood, 1998). Wave V representing the neural activity in
the lateral lemniscus and/or inferior colliculus (Hood, 1998).
3
2.2 Stimulus factor
Both latency and amplitude of the early evoked potentials can be affected by
manipulating the stimulus. Thus, it is necessary to obtain the best possible response and
correctly interpret test result by using an optimal stimulus parameter.
2.2.1 Type of stimulus
There are various types of stimulus used in ABR recording. ABR is best elicited
by stimulus with brief onset due to its high dependency to the neural synchrony.
Therefore, clicks are favored as a stimulus because it nature of abrupt onset and
broadband spectrum that can elicit good neural synchrony at broader region of
frequencies in basilar membrane (BM) and make it produce a robust response when
measuring from the scalp (Katz, 2002). The neural synchrony occur in BM is based on
the cochlear onset neuron in auditory nerve. Onset chopper (Oc) neurons have very
specialized membrane properties and precise temporal processing. Hence, Oc neurons
exhibit a wide dynamic range and robust firing to broadband stimuli and it suggested that
they may have a role involved in signal processing in noise and in the detection of
spectral cues related to sound localization (Mulders et al., 2007).
Tone burst is another type of stimulus with brief onset that has specific frequency
where it activates the restricted part of the basilar membrane in the cochlea by using
certain stimulus envelope. However, due to its fast rise and fall time it has high tendency
to have ‘spectral splatter’ effect especially in low frequency tone burst that causing of
unwanted contribution from high frequency region of the BM (Hall, 1992).
4
2.2.2 Stimulus polarity
Polarity is referred to the direction or movement of transducer diaphragm in
relation to the pressure wave generating at tympanic membrane. Condensation,
rarefaction and alternating are three categories of stimulus polarity. Condensation
polarity occurs when movement of transducer’s diaphragm is towards the tympanic
membrane producing positive direction of sound wave, whereas movement of transducer
away from tympanic membrane is called rarefaction polarity (negative direction).
Alternating polarity is switching mode between both two polarities (Hall, 1992).
Garga et al. (1991) reported that polarity will affect latencies for stimulus
conditions response dominated by low frequency energy whereas at the high frequency
energy there are no such affect were observed. These finding are completely consistent
with behavior of individual hair cells and neurons within the auditory pathway true for
normal hearing. According to Fowler et al. (2002), rarefaction clicks are expected to
produce shorter latencies and greater amplitude for ABR compared to condensation. In
contrast, Stockard et al. (1979) noted that 15% to 30% of normal subjects may show the
opposite polarity pattern where shorter latency values for condensation than for
rarefaction clicks (Hall, 1992).
Even though there are arguments whether to use rarefaction or condensation
stimuli. Study showed wave V amplitude tends to be larger in response to condensation
stimuli but there is no significant latency difference in wave V latency to rarefaction or
condensation stimuli in normal hearing individuals (Hood, 1998). This study used
5
condensation stimuli. It is based on presence or absence of wave V with subjective
detection to identify the time.
2.2.3 Stimulus intensity
The site of ABR generation along the basilar membrane is related to intensity. In
general principle, ABR latency decreases and amplitude increases with greater stimulus
intensity. There are two reasons behind it. Firstly, it is due to progressively faster rising
generator potential within the cochlea causing similarly faster development of excitatory
postsynaptic potentials (EPSPs). Besides, shorter travel time from the oval window to
basal end occur if use high stimulus level resulting shorter latency of ABR waveform
(Hall, 1992). This was supported by Picton et al. (1981) where high stimulus level
between 75 dBnHL to 95 dBnHL may activate the basal part and then moves
progressively toward apex for lower intensity level when decrease from 70 to 80 dBnHL
(Hall, 1992). Only wave V is clearly visible whereas the earlier component tends to
become indistinguishable at 35 dBnHL (Hood, 1998).
2.2.4 Stimulus rate
Stimulus rate is defined as number of stimulus presented in one seconds. Rate
need to be presented more than the duration of ABR which are more than 15 milliseconds
(Hall, 1992). In general principle, stimulus repetition rates up to approximately 20 clicks
per seconds have little effect on the ABR. However, above that rate the latency will
increase and amplitude decreases (Burkard et al., 2007; Hood, 1998).
Each wave component has different effect of ABR amplitude toward stimulus
rate. Wave V amplitude show less decrement with increase of rate from relatively slow
6
rate (8 to 33/sec) to rapid rate (80 to 90/sec) than the earlier component. At higher rate,
amplitude for wave V typically less than about 10% to 30% from the amplitude of slower
rate while wave I amplitude is less about 50% (Hall, 1992). Thus, less testing time
achieved when using higher stimulus rate that permit large collection of data for clinical
threshold estimation. Besides, by increasing repetition rate it will cause latencies shift due
to adaptation of neuron which does not have enough time to recover (Don et al., 1977;
Pratt & Sohmer, 1976).
Stimulus rate may have different effect based on age. Prolongation of ABR
latency is more prominent for younger children (under age of 18 months) than adult
(Stockard et al., 1977). Consequently, the infant ABR is characterized by delayed
interwave intervals. For instance, an interwave interval for normal newborn in average is
about 5 ms compared to 4 ms in adult. In anatomy and physiology of central nervous
system (CNS), delayed interwave latencies is due to incomplete nerve fiber
myelinization, reduced axon diameter and immature synaptic functioning which cause
prolonged neural transmission in younger subjects (Hall, 1992).
Whilst high rate may affect the quality of ABR waveforms, it is reported that high
stimulus rate is useful to determine site of lesion testing (Hall, 1992) as well as the ability
to reduce ABR acquisition time (Eysholdt & Schreiner, 1982). However, high stimulus
rate cause time between presentations of stimuli equals to limit of ABR conventional
averaging responses about 20 ms. Thus, at rates faster than limit, responses will overlap
and interfere to one another and causing distorted and useless final averaging.
Interactions between responses also take place if high stimulus rate is applied to second-
order or third-order neuron (Burkard et al., 2007).
7
In conventional ABR the presentation of click stimulus can only take place when
response to the previous stimulus is over (Eysholdt & Schreiner, 1982). If stimulus is
being presented before the response of previous stimulus is obtained, those ABR waves
will be overlapped to each other causing ABR signal will be deteriorated. This could be
problematic for audiologist to interpret ABR waveforms. Although there is some
possibility reduction of ABR amplitude and increase wave latencies, those effect does not
hinder the ability for audiologist to detect wave V ABR.
2.3 Maximum Length Sequences (MLS)
Maximum Length Sequences (MLS) is pseudorandom binary sequences first
introduced to check feasibility of obtaining response and speculated that it will lead to
improvement of recording technique with reduce time taken to find average (Eysholdt &
Schreiner, 1982). It works by process of deconvolution where response recorded using
MLS stimulation can be extracted from the multiply overlapped combination of
responses that will be generated by the stimulus sequence. MLS can overcome several
problem occur when high stimulus rate was used.
Firstly, MLS has special mathematical properties which intervals between stimuli
vary in quasi-random fashion where ABR responses that overlap can be compensated
causing faster stimulus rate can be used. Hence, waveform obtained at lower rate from
conventional averaging can be extracted from MLS stimulation at high rates.
Secondly, MLS stimulation with proper decoding software can cause response
interactions of stimulus similar to the conventional response. Besides, MLS stimulation
also reveals nonlinear temporal interaction between responses of human auditory system
8
(Shi & Hecox, 1991). A nonlinear interaction occurs between responses when the time
between presentations of stimuli is small and stimulus rate permitted by MLS technique
only. The properties of the nonlinear components differ from those of the conventional
component and there is preliminary evidence that they may provide a measure that is
more sensitive to pathology than the conventional component (Burkard et al., 2007).
Previous studies showed that high stimulus repetition rate through MLS generally
can improve the ABR testing time compared to the conventional stimulus rate via
different technique to identify ABR wave (Eysholdt & Schreiner, 1982; Leung et al.,
1998; Thornton & Slaven, 1993). The fixed number of sweeps technique use for MLS in
ABR wave detection theoretically can decreased test time by a factor of 8 in linear MLS
stimulation (Eysholdt & Schreiner, 1982). Same goes to the technique that used MLS
with SNR. The speed of test relative to the conventional rate of 33.3 cps, calculated using
the formula c2.k2. (r/r0) (Thornton & Slaven, 1993). This is the speed which the test can be
carried out when averaging to the same SNR and takes into account the reduction in wave
V amplitude seen with increasing rate and decreasing stimulus level. At linear MLS of
300 cps the relative speed of test is greater than all stimulus repetition rates level in
conventional ABR and MLS. As the rate increases and the level decreases, the relative
speed of test also decreases, until by 1000 cps testing to the same SNR takes longer than
the conventional case at all stimulus levels. Thus, the effective test time using MLS with
SNR is between 200 cps till 300 cps (Leung et. al., 1998).
According to Hall (1992) there is also statistical and clinically significant
relationship between rates of transient stimuli with behavioral auditory threshold. From a
stimulus rate of 5 cps to 80cps, threshold is enhanced by 5 dBpeSPL. This is due to large
9
temporal summation of acoustic energy. However by increasing the stimulus repetition
rates up to 1000 cps, the ABR waves cause longer wave latencies and ABR waveform
amplitude reduced due to incomplete recovery of auditory system from previous and
continuous stimulation (Eysholdt & Schreiner, 1982; Burkard et al., 1991; Burkard et al.,
2007). However, changes are not the same for each wave component. Even though
amplitude for wave V has typically decreased about 10% to 30% relative from the
amplitude of slower rate; it is acceptable to use high repetition rates because of clear
wave V ABR morphology remains there.
Besides, MLS also feasible to test infant ear binaurally. According to Hood
(1998) study done to compare between standard ABR and MLS ABR at several
intensities for normal infant showed less than 5 minutes total test time needed for click
stimuli on both ears with the MLS technique.
Researchers also have been exploring the possible MLS applications in various
areas. In the study of ABR, it has been noticed that reliable ABR were produced
remarkably comparable to conventional ones in morphology (Burkard et al., 1991).
During recording ABR of premature infants by MLS, Weber & Roush (1993) found that
the clarity of ABR could be well-defined under a rate as high as 900 cps and the quality
of MLS ABR even better than conventional ABR especially in large noise environment.
It is suggested that MLS could be applied in newborn hearing screening. This suggestion
was verified by Jiang et al. (1999). Besides, they also employed MLS in asphyxiated
neonates and found that central auditory impairment of these neonates was more
detectable with MLS paradigm (Jiang et al., 2001)
10
2.3.1 MLS limitations
Despite the advantages of MLS to reduce the ABR test time; there were
arguments that MLS ABR is intrinsically noisier than conventional ABR with equal
number of stimuli (Marsh, 1990; Burkard, 1991; Picton et al., 1992). Marsh (1992)
reported that for each MLS condition compared to conventional ABR, wave V amplitude
decreased more than the noise level supporting the argument of a reduced signal to noise
ratio (SNR). Thus, with a poor signal to noise ratio none of the MLS conditions was as
efficient as the conventional method. Besides, issues regarding low SNR obviously affect
ability to identify presence and reliability of response in ABR testing (Lasky, Perlman &
Hecox, 1992). Despite this, the quality of MLS ABR waveforms was reported to be better
than conventional ABR based on audiologist report even with poor SNR (Weber &
Roush, 1995). Three judges evaluated the quality of ABR responses and among fifty
newborns, 32% had more clearly defined ABR responses obtained with MLS compared
to conventional ABR.
Burkard (1991) noted that more number of MLS pulses must be presented in
order to achieve same SNR as the conventional response recorded at slower rate. The
number of additional pulse must be presented equal to amplitude of MLS response and
amplitude of conventional response (Lasky, Shi & Hecox, 1992).
There are several factors need to consider on how MLS allows greater rate than
100 cps might reduce the time required to obtain constant SNR response with application
of more number of MLS pulses. Firstly, the increased rate must more than offset the
11
decrease in response amplitude. Thus, amplitude of the wave V must be reduced less than
a factor of 1.414 for each doubling in rate. This is assuming that SNR increases in
proportion to the square root of the number of sweeps. Secondly, there is a 3 dB loss in
SNR with use of cross correlation procedure. Basically, the cross correlation procedure
results in a summation of noise over almost twice as many events as the number of
stimulus presentations. Hence, this result in a loss of 3 dB in SNR and roughly twice as
many stimuli are necessary to recover this loss in SNR. Other than that, at very
sufficiently high rates the noise epochs are not truly independent. If these noise epochs
are partially correlated, SNR will not recover in proportion to the square root of the
number of sweeps but at some lesser value (Burkard et al., 2007). Lastly, although that is
not measurable in the ABR, it is unlikely that the ABR for each stimulus in the train is
identical. This is because the interstimulus interval is not constant in train. It is likely that
the ABR reaches an adapted state after a few stimuli. Based on Don et al. (1978) when
clicks are presented in trains the ABR adapts following 3 to 5 clicks and no further
changes in the ABR are noted for responses to additional clicks in the train. It maybe the
reason of MLS ABR looks like conventional ABR. However, it probably there is still
some recovery in the response if it is not deterministic which means presentation of each
stimulus is not similar.
Even though improvement in SNR depends on the response being the same for
each stimulus presentations, the nondeterministic nature of the ABR due to temporal jitter
in MLS stimulus presentation will also reduce the rate of SNR improvement across the
number of stimulus presentations. To date, no one has included all of those factors in
determining whether the MLS procedure allows collecting ABR more efficiently.
12
2.4 Waveform detection
Waveform detection is referring to the time taken to determine whether an ABR
waveform was present or absent whereas waveform analysis is referring to the time taken
to both detect and analysis the ABR waveform. There are two methods that can be used
for ABR detection which are objective and audiologist detection (Ahmad et al., 2008).
2.4.1 Audiologist detection
Audiologist detection is a technique where the presence of ABR waveform is
made by clinician upon visual inspection of the recorded waveforms under favourable
conditions (Arnold, 1985; Don, Elberling & Waring, 1984). The criteria of visual
judgment includes clear ABR component in the waveform as well as waveform
repeatability. By using audiologist detection, human operator has full authority over the
waveform interpretation. Thus, all relevant considerations must be taken into account
before making any diagnosis for each case. Nevertheless, at least two problems occur
with audiologist detection. Firstly, tester will take a longer time to observe and visualize
many waveforms for each patient when performing this technique (Hall, 1992) as well as
the possibility of error rises when too many waveforms must be interpreted within a short
period of time. Secondly, there is the possibility of mis-labelling and errors during the
waveform interpretation, especially with persons who have less experience at interpreting
an ABR waveform (Ahmad et al., 2008).Based on Elberling & Don (2007), the methods
13
used for audiologist detection can be divided into four categories which are response
judgment made by one observer, response replication, response tracking, and response
judgment by multiple observers.
2.4.1.1 Response judgment made by one observer
In this technique, human operator use his or her experience to interpret whether or
not there is an ABR based on the peaks and latency in the waveform, It usually applies to
condition where the ABR waves can be seen clearly at high intensity (more than 70
dBnHL) (Ahmad et al, 2008). This technique is considers fast because it only require
single run. However, large amount of noise sometimes produce peaks within the usual
latency of a normal ABR waveform which can contribute to the tendency for a false
positive result (Ahmad et al, 2008).
2.4.1.2 Response judgment by multiple observers
In this technique, more than one observer will interpret the ABR response which
aims to reduce the errors made when a single observer is used. The validity for this
technique is based on the assumption that all observers are independent (Ahmad et al,
2008). However, it is not always true; in some cases where both observers have similar
training procedure for ABR waveform interpretation there is a possibility that both of
them could make a false positive interpretation or a similar judgment even if both
observers are independent (Ahmad et al, 2008). The reliability made by Kappa test also
showed that there was considerable disagreement between judges in their decisions in
determining the presence of ABR. Thus, subjective detection by clinician may introduce
bias and inaccurate ABR interpretation (Arnold, 1985).
14
2.4.1.3 Response replication
In this technique, two ABR averages are recorded and overlayed on each other.
The degree of overlap between the two traces is observed and compared. If there is
consistent overlap between those two traces particularly of peaks occurring at expected
latencies and amplitudes, it is more likely that an ABR is present in the waveforms. The
principle in this technique is that the ABR will replicate and the random noise will
fluctuate over time. However, it might be a problem if the noise in two averaged differs
in large amount causing poorer replication and difficult interpretation. Besides, this
approach takes time because it requires extra test runs to acquire for each original and
repeated waveform. (Ahmad et al, 2008).
2.4.1.4 Response tracking
In response tracking, the waveforms are observed as a function of the stimulus
intensity usually done either by one or multiple observers. This technique allow observer
to review the response based on the expected changes in the waveforms, such as a
decrease in amplitude and an increase in the latency, while he or she decreases the
stimulus intensity. For instance, clinician may start testing at where the ABR peaks are
often clearly observed (80dBnHL), and then reduce the intensity and track the expected
changes in the Wave V of each new waveform. When no peak was obtained at certain
intensity testing can be stopped. Clinician need to repeat the waveforms at any intensity
to ensure his or her judgment is true. By using high intensity waveforms it can be
template for the low intensity waveforms by observing the changes in amplitude and
latency. However, problems with this technique include the effects of noise on the
15
expected changes in the ABR waveform, and the time required to complete the multiple
recording runs (Ahmad et al, 2008).
2.5 Conclusion
Although MLS have limitation which is intrinsically noisier than conventional
ABR. It can be solved out by using more number of MLS pulses presented in one
seconds in order to achieve same SNR as the conventional response recorded at slower
rate. In previous study, there are many techniques can be applied for MLS ABR to reduce
the test time in detection of ABR wave whether by fixed number of sweeps or SNR.
However, all of those study only concern regarding the conventional ABR and linear
MLS. Therefore, this current study investigated different technique which is comparing
the ability of linear and nonlinear MLS with combination of audiologist detection with
the aim to decrease testing time in infant hearing screening.
16
2.6 General objective
The purpose of this study is to investigate the improvement of ABR testing time
between high stimulus repetition rates and slow stimulus repetition rates with
combination of audiologist detection (waveform analysis).
2.6.1 Specific objectives
i. To compare the time for detection of ABR waveform between standard
click and linear MLS with audiologist detection.
ii. To compare the time for detection of ABR waveform between linear MLS
and non linear MLS with audiologist detection.
iii. To compare the time for detection of ABR waveform between standard
click and non linear MLS with audiologist detection.
17
Chapter III
METHODOLOGY
3.1 Study design
Quasi-experimental design was used for this research. This study design involves
selecting groups which variable is tested without any random pre selection process. For
example, there is no randomization in term of pre-determined effects of gender, age and
hearing thresholds on subject performance. After the selection, experiment was preceded
in a similar ways to any other experiment where variable being compared between
different groups (Shuttleworth M., 2008). In this study, the time to detect ABR waveform
using different stimulus repetition rate applied to the same subject will be investigated.
The data collected in this study were obtained from the same subject who are exposed to
different stimulation or often called as repeated measures analysis. It is useful because
individual differences can be eliminated since the data was collected from the same
subject under repeated conditions. Besides, sample is not divided between conditions
such as gender, age and study population. Hence, inferential testing become more
powerful (Choudhury A., 2009).
3.2 Equipment/Software
Compaq Presario V3000
Matrix Laboratory (MATLAB) software
Statistical Package for Social Sciences (SPSS) Version 16.0 software.
18
3.3 Study place
International Islamic University Malaysia (IIUM), Kuantan, Pahang Campus.
3.4 Study population
3.4.1 Population
Subject of this study was sampled from newborn population involved in Newborn
Hearing Screening at Southern East of Queensland, Australia. This study was cleared by
Matter Health Services Human Research Ethics Committee (Clearance number: 770_mc)
and clear by ethics committee of School of Health and Rehabilitation Sciences University
of Queensland (Protocol number: A517/..kr). Besides, Kulliyyah of Medicine (KOM)
Ethics Committee from International Islamic University Malaysia (IIUM) also approved
the ethics application for this study
3.4.2 Subject selection criteria
In previous study, there are 144 subjects were involved. From the overall data, not
all of data contain good and clear ABR waveform. Thus, 20 out of 144 subjects were
selected based on criteria that had been met. The criteria of data being selected were
based on clear ABR wave as well as waves are not contaminated by muscle artifact and
electroencephalogram (EEG). These criteria were set in order to prevent from faulty
result and test validity and reliability.
19
3.4.3 Subject description
20 subjects were newborns data that pass Neonatal Hearing Screening at Southern
East of Queensland, Australia. From that, there are 15 males and 5 females involved in
this recent offline data analysis.
3.5 Sampling technique
No sampling was done since the data was already collected from previous
Dzulkarnain A.A., (2008). However the data collected from this study was sampled using
convenience sampling technique. In conventional sampling technique the researcher
recruited their subject based on their availability. During the data collection, researcher
recruited the subject by approaching their primary care giver whether they are interested
to be a part of the research subject’s on top of their compulsory Newborn Hearing
Screening using AABR. The actual AABR test procedure was summarized in Appendix
1.1
3.6 Procedures
1 Ahmad et al. (2008). The method to improve maximum length sequences auditory brainstem response analysis and
detection time. Audio Neurotol, 13,
20
Since data has been collected from previous study, the aim of this study is to use
the same data for different offline analysis. It was conducted using custom code written
using Matrix Laboratory (MATLAB). The data is in the forms of binary M-files contains
of ABR waveform recorded at 35 dBnHL at six different stimulus repetition rates
(standard clicks: 33 & 90 clicks per second (cps), linear and nonlinear MLS: 180, 250,
500 and 833 cps). This MLS repetition rates were used because those rates represent
slightly higher rate, middle higher rate and extreme rate. However, this study used an
arbitrary rates rather than following previous literature repetition rates in order to see if
slight change to the typical rate numbers will give some effects. This further supported by
good result obtained previously (Ahmad Aidil, 2008). Therefore, the following
procedures were applied:
1. 20 data of subjects that had met the criteria were selected.
2. Then, case study was conducted in order to determine the best parameter to use in
term of type of averaging, type of artifact rejection, filter setting and comparison
setting. Hence, all criteria were finalized before the actual data analysis was
conducted. The result of case study used mean averaging, 15 μv for artifact
rejection, filter setting (low frequency cut filter 100Hz and high frequency cut
filter 5kHz) and comparison setting by elapsed time subplots.(See Appendix II)
3. MATLAB software is opened. Then, ABRpublish is entered. Once the browse
button appeared, Materdatafile folder was selected to search for subject number
that fulfills the criteria. Parameter settings from the case study were set first. Next,
data subjects of ABR wave were plotted from zero seconds to the last seconds of
21
each recording for standard click (click 33cps and click 90cps) linear and
nonlinear MLS at different stimulus repetition rates (180cps, 250cps, 500cps and
833cps). After that, the time for ABR wave’s detection was determined by the
researcher. Those times to detection and ABR waves were given to an audiologist
to confirm the earliest time that an ABR can be detected for each stimulus
repetition rates.
3.7 Waveform detection criteria
There are widespread of technique in identifying ABR. In this study, a response
replication technique was used. In this technique, the area of overlapping between traces
and the degree of replication between traces are used as the criterion to detect the
presence of ABR waveforms. Firstly, the time detection of wave V is made by the
researcher. After that, all the data were saved and confirmed by an audiologist. Thus, for
the final data analysis, we used the time or data that was obtained from an audiologist.
The audiologist who confirmed the time to detection for ABR wave has more than 7
years experience in clinical and auditory electrophysiology field especially in ABR
3.7 Conclusion
22
In overall, by using waveform detection criteria researcher need to figure out the
testing time between three groups of stimulus repetitions rates which are in standard click
(33cps and 90cps) as well as linear and non linear MLS (180cps, 250cps, 500cps and
833cps).
3.8 Data analysis
3.8.1 Variables
The main dependent variable in this study is the time to detection and the
independent variables are the ABR with various stimulus rate. The stimulus rate include
standard clicks is 33 and 90 clicks per seconds and both linear and non linear MLS at
180, 250, 500 and 833 clicks per seconds. The independent variables are something that
can be changed by experimenter in order to do the actual experiment. The independent
variable for this study is the type of stimulus and stimulus rate. The dependent variable is
the outcome of independent variables when it changes. In this study the dependent
variable is time taken to detect ABR waveform using subjective detection. Table 3.1
shows the summary of dependent and independent variables in this study.
Table 3.1 Summary of dependent and independent variables
23
________________________________________________________________________Dependent variable Independent variable Types________________________________________________________________________Time taken to detect 1. Type of stimulus/algorithm Standard click, linear and ABR waveform nonlinear MLS
2. Types of stimulus repetition #Click 33 cps and 90 cps rates #Linear and nonlinear MLS
180cps#Linear and nonlinear MLS 250cps#Linear and nonlinear MLS 500cps#Linear and nonlinear MLS 833cps
Weber, B.A., Roush, P.A. (1993). Application of maximum length sequence analysis to
auditory brainstem response testing of premature newborns. Journal Am Acad
Audiol, 4, 157-162.
51
APPENDIX 1: AABR TEST PROCEDURE
AABR test procedure2
Each subject was assessed by the researcher using the PC-based AABR device,
after he or she had been assessed by the staff of the ‘Healthy Hearing: Newborn Hearing
Screening’ program using their AABR device. The staff of the ‘Healthy Hearing:
Newborn Hearing Screening’ program performed their AABR assessment on each
newborn by:
1. Preparing three sites on each newborn’s scalp and shoulder using alcohol wipes
and a skin-preparation solution. These sites were vertex (Cz), the nape of the
neck and one shoulder.
2. Placing their surface electrodes (jelly tab® from Natus) on these sites and
connected these electrodes to their AABR device. The impedance of these
electrodes was kept below 5 kOhms.
3. Positioning their earphones (flexicoupler® from Natus) from their AABR
device over the newborn’s ears
4. Initiating the test sequence on their AABR device, the ALGO 3® (from Natus)
5. Waiting for their AABR test to finish.
The researcher then completed his AABR assessment on each newborn by:
6. Leaving the electrodes from the above testing on the newborn
2
52
7. Disconnecting the hospital AABR device from these electrodes and connected
PC-based AEP device to these electrodes
8. Removing the hospital’s earphones and inserted the PC-based AABR device’s
insert earphones into one of the newborn’s ears only (whatever ear was most
accessible at the time of testing)
9. Entering the newborn’s research number into the PC-based AEP device
10. Initiating the test sequence on his PC-based AEP device
11. Waiting for the PC-based AEP device to finish its test run (this took
approximately eight minutes per newborn)
12. Disconnecting the newborn from the PC-based AEP device and returning the
newborn to the care of his or her parents/caregivers
13. Recording the newborn’s date of birth, date of test, gender, and relevant medical
history including any risk factors for hearing loss.
53
APPENDIX 2: AABR STIMULUS AND RECORDING PARAMETERS
AABR stimulus and recording parameters
Tables 1.1 and 1.2 shows the stimulus and recording parameters, respectively, used in the
previous study by Ahmad et al. (2008):
Table 1.1: The stimulus parameters used in previous study
Stimulus Parameter
PC-based AEP System
Intensity level 35 dBnHL
Transducer EAR-3A insert phones
Test ear Right or left (based on which ever ear was most accessible at the time of the testing)
Type Two stimulus types were used separately:
An acoustic click driven by a 0.1 ms electrical click in a conventional presentation sequence.
An acoustic click driven by a 0.1 ms electrical click in a MLS of order six, consisting of 32 clicks and 31 silences.
Polarity Condensating
Rate Two stimulus rates were used separately for the conventional stimulus: 33 and 90 cps.
Four maximum stimulus rates were used separately for the MLS stimulus: 180, 250, 500, and 833 cps.
MISI The MISI for the 4 MLS stimuli listed above were 5.56, 4, 2, and 1.2 ms respectively
54
Table 1.2: The recording parameters used in previous study
Recording Parameter
PC-based AEP System
MLS reconstruction technique
Two MLS reconstruction techniques were used:
linear, bipolar and linear, unipolar
Gain 30000 (A/D converter voltage of ±5 V with a 32-bit resolution)
Electrode montages
Vertical
Analysis window
0 to 15 ms for all stimuli except the conventional stimuli at 90 cps, which had an analysis window of 0 to 11.13 ms.
Sampling frequency
32 890 Hz
Low frequency cut filter
100 Hz
High frequency cut filter
5000 Hz (filtered off line to 3000 Hz)
Number of averages
2560 for each conventional stimulus
5120 for the 180 cps s maximum stimulus repetition rate (the MLS stimulus was presented 160 times), 7680 for the 250 cps rate (the MLS stimulus was presented 240 times), 8960 for the 500 cps rate (the MLS was presented 280 times), and 10240 for the 833 cps rate (the MLS stimulus was presented 320 times)
Masking Nil
Artifact rejection threshold
15 µV
Note: The vertical electrode montage required the non-inverting electrode to be placed on the vertex, the inverting electrode to be placed on the back of the neck, and the ground electrode to be placed on either shoulder.
55
APPENDIX 3: DATA COLLECTION RESULTS FOR ABR WAVEFORM DETECTION TIME USING MATLAB SOFTWARE