Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effect of noise floor on cortical auditory evoked potential response detection at threshold intensities Ritchie, Heather 2019

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2019_november_ritchie_heather.pdf [ 1.36MB ]
Metadata
JSON: 24-1.0384598.json
JSON-LD: 24-1.0384598-ld.json
RDF/XML (Pretty): 24-1.0384598-rdf.xml
RDF/JSON: 24-1.0384598-rdf.json
Turtle: 24-1.0384598-turtle.txt
N-Triples: 24-1.0384598-rdf-ntriples.txt
Original Record: 24-1.0384598-source.json
Full Text
24-1.0384598-fulltext.txt
Citation
24-1.0384598.ris

Full Text

  THE EFFECT OF NOISE FLOOR ON CORTICAL AUDITORY EVOKED POTENTIAL  RESPONSE DETECTION AT THRESHOLD INTENSITIES   by  Heather Ritchie  B.H.Sc. University of Ottawa, 2014   A THESIS SUBMITTED IN PARTIAL FULFULLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Audiology and Speech Sciences)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  October 2019  © Heather Ritchie, 2019   ii  Supervisor and Committee The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, a thesis entitled:  The effect of noise floor on cortical auditory evoked potential response detection at threshold intensities     submitted by Heather Ritchie  in partial fulfillment of the requirements for the degree of Master of Science in Audiology and Speech Sciences  Examining Committee: Dr. Anthony Herdman, Professor and Electrophysiologist Supervisor  Sasha Brown, Professor and Clinical Audiologist Supervisory Committee Member  Dr. Navid Shahnaz, Professor and Clinical Audiologist Supervisory Committee Member           iii  Abstract  Cortical auditory evoked potentials (CAEPs) are neural responses that occur in response to changes in sound, which can be recorded from electrodes placed on the scalp. The N1-P2 response can be reliably used to determine a person’s frequency-specific hearing thresholds. This objective method for hearing assessment is used clinically in situations where the patient is unable or unwilling to provide reliable behavioural responses. Currently, the gold standard method of interpreting CAEP results is dependent upon the visual judgment of the clinician. A high level of noise in the recordings may obscure a small N1-P2 response. Established criteria for an acceptable residual noise (RN) level does not currently exist. Such criteria could be used as a tool to assist in the interpretation of CAEPs, making judgments more reliable among clinicians. The goal of the present study was to estimate a noise criterion based on recorded and simulated CAEP averages with various different levels of RN. CAEP results at threshold and sub-threshold intensities were recorded for 12 normal hearing adults at 500 Hz and 2000 Hz. Each waveform was presented a total of five times, with each average including a different number of sweeps. Simulated CAEPs were generated (N=37), with averages including a variety of different RN levels. The waveforms were presented to three expert raters, who judged each waveform independently as having an N1-P2 response present or absent.  RN criterion, maximal noise floor allowable for judging a response as truly absent, was determined when raters performed at 95% sensitivity of detecting a true response. The recommended criteria for the maximum noise floor level were 0.111 μV (relative to the pre-stimulus interval) or 0.145 μV (relative to the post-stimulus interval). Simulated data exhibited a iv  better face validity than the recorded CAEPs. Further studies may lead to the implementation of new guidelines surrounding RN criteria in CAEP recording procedures. Such guidelines may provide more validity and reliability in clinical practice.                         v  Lay Summary  Objective measures of hearing ability must be used in cases where an individual is unable or unwilling to provide reliable responses behaviourally. Clinicians can record cortical auditory evoked potentials (CAEPs), which are brain waves that occur in response to sound. If a sound is heard, a response is visible within the CAEP. Smaller responses may be obscured by background EEG noise. A target level of residual noise (RN) can help clinicians to determine whether or not a response is present. This value for CAEPs does not currently exist. This study analyzed the noise levels present in CAEPs close to hearing thresholds based on the judgments of three independent raters. If the noise floor is above the recommended RN criterion of 0.111 μV (pre-stimulus RN) or 0.145 μV (post-stimulus RN), a small response could be obscured by background noise. Further studies may lead to the implementation of new RN guidelines.             vi  Preface  This dissertation is an original work of the author H. Ritchie and thesis supervisor Dr. Anthony T. Herdman. The study design, data collection, and analysis of research data were conducted in the BRANE laboratory at the University of British Columbia. The methods were approved by the Behavioural Research Ethics Board of the University of British Columbia. The ethics certificate number is H14-00441, under the project title of “CAEP Gap Study.”                     vii  Table of Contents  Abstract .......................................................................................................................................... iii Lay Summary .................................................................................................................................. v Preface............................................................................................................................................ vi Table of Contents .......................................................................................................................... vii List of Tables ................................................................................................................................. ix List of Figures ................................................................................................................................. x List of Abbreviations ..................................................................................................................... xi Acknowledgements ....................................................................................................................... xii Dedication .................................................................................................................................... xiii Chapter 1: Introduction ................................................................................................................... 1 1.1 Cortical Auditory Evoked Potentials and Hearing Threshold Estimation ............................ 1 1.2 Subject-related Factors .......................................................................................................... 3 1.3 Stimulus-related Factors ........................................................................................................ 5 1.4 Response Detection ............................................................................................................... 7 1.4.1 Background Noise .......................................................................................................... 7 1.4.2 Averaging ....................................................................................................................... 8 1.4.3. Methods of Response Classification .............................................................................. 8 1.5 Goals of Study ..................................................................................................................... 10 Chapter 2: Methods ....................................................................................................................... 12 2.1 Participants .......................................................................................................................... 12 2.2 General Study Procedures ................................................................................................... 12 2.2.1 Audiometric Assessment .............................................................................................. 12 2.2.2 Electrophysiological Measures ..................................................................................... 13 viii  2.3 CAEP Rating Procedure ...................................................................................................... 14 2.4 Simulated CAEP Data ......................................................................................................... 16 2.5 Simulated CAEP Data Rating Procedure ............................................................................ 17 2.6 Statistical Analysis .............................................................................................................. 17 Chapter 3: Results ......................................................................................................................... 19 3.1 Recorded and Simulated CAEP Results.............................................................................. 19 3.2 Rater Judgments for Recorded and Simulated CAEPs ....................................................... 22 3.3 Inter-Rater Reliability ......................................................................................................... 26 3.4 Sensitivity Index and Response Bias .................................................................................. 27 3.5 Pre-stimulus and Post-stimulus Noise Levels ..................................................................... 27 Chapter 4: Discussion ................................................................................................................... 29 4.1 CAEPs and Noise Level Measures...................................................................................... 29 4.1.1 Recorded CAEPs .......................................................................................................... 30 4.1.2 Simulated CAEPs ......................................................................................................... 30 4.1.3 Pre-stimulus and Post-stimulus Noise Level Measures ................................................ 31 4.1.4 Selection of an RN Level Criterion .............................................................................. 31 4.2 Inter-rater Reliability ........................................................................................................... 32 4.3 Caveats ................................................................................................................................ 33 4.4 Clinical Implications and Directions for Future Research .................................................. 34 4.5 Conclusions .................................................................................................................... 35 Bibliography ................................................................................................................................. 36    ix  List of Tables  Table 3.1: Inter-rater Reliability for all CAEP Judgments ........................................................... 26 Table 3.2: Sensitivity Index and Response Bias for Recorded and Simulated CAEPs ................ 27 Table 3.3: Noise Levels for True Positive Judgements at 95% Sensitivity .................................. 28                       x  List of Figures  Figure 2.1: Format for Rating CAEP Thresholds ......................................................................... 15 Figure 3.1: Recorded CAEP Results ............................................................................................. 20 Figure 3.2: Simulated CAEP Results ............................................................................................ 21 Figure 3.3: Rater 1 Judgments for Recorded and Simulated CAEPs ........................................... 23 Figure 3.4: Rater 2 Judgments for Recorded and Simulated CAEPs ........................................... 24 Figure 3.5: Rater 3 Judgments for Simulated CAEPs................................................................... 25                     xi  List of Abbreviations  Abbreviation   Definition   ABR    Auditory Brainstem Response BCEHP   British Columbia Early Hearing Program CAEP    Cortical Auditory Evoked Potential EEG    Electroencephalography IHS    Intelligent Hearing System  ISI    Inter-stimulus Interval RN    Residual Noise  SNR    Signal to Noise Ratio                  xii  Acknowledgements First and foremost, I would like to express my deepest gratitude to my thesis supervisor, Dr. Anthony Herdman, for his guidance, expertise, and enthusiasm. Thank you for dedicating your time to my thesis project, and for making the research process interesting and enjoyable. I would also like to thank my thesis committee, Sasha Brown and Dr. Navid Shahnaz for their invaluable guidance, input, and support. A special thank you to the members of the BRANE Lab: AJ Hildebrand, Natalie Tran, Seho Bann, and Nima Touss. Thank you all for your encouragement and support throughout the research process.  Next, I would like to thank my former classmate and colleague Stephanie Strahm for her endless encouragement and reassurance. Finally, thank you to Matthew Hegmann for being a constant source of comfort and motivation.          xiii   Dedication   To my family: Your love and support have enabled me to accomplish more than I had imagined. Thank you for inspiring me to pursue my dreams.            1  Chapter 1: Introduction 1.1 Cortical Auditory Evoked Potentials and Hearing Threshold Estimation Cortical Auditory Evoked Potentials (CAEPs) are electrical changes in neural activity in response to an auditory stimulus. Responses are recorded via electroencephalography (EEG) with electrodes placed on the scalp. A P1-N1-P2 waveform occurs in response to the onset or offset of an auditory stimulus (Davis, 1939). The response may be evoked by changes in the frequency, intensity, and temporal properties of a sound (Jones, Longe & Vaz Pato. 1998). The mature P1-N1-P2 response observed in adult populations is comprised of three components: a positive peak (P1), a negative peak (N1), and a second positive peak (P2) (Näätänen & Picton, 1987). The latencies of these waveforms are 50 ms, 100 ms, and 170-200 ms, respectively (Davis, Mast, Yoshie, & Zerlin, 1966). Peaks correspond with the activation of different cortical regions as the acoustic information ascends toward the auditory cortex and areas of higher processing (Picton, 2010).  In adult populations, these neural generators include several regions predominantly located in the primary auditory cortex and auditory association areas (Näätänen & Picton, 1987). As P1 is not always identifiable, the N1-P2 amplitude is seen as the most reliable method of measuring the magnitude of a response across test subjects (Davis & Zerlin, 1966). The N1-P2 response complex can be used to measure the physiological auditory capacity of the participant (Van Maanen & Stapells, 2005). CAEP thresholds are defined as the lowest stimulus intensity where a replicable response can be obtained (Vanaja, 2016). A present N1-P2 response that is time-locked to the onset of an acoustic stimulus indicates that the information has been detected by the auditory cortex of the participant (Lightfoot, 2016). This indicates that a synchronous neural activation in the segment of the central auditory system between the cochlea and auditory cortex has occurred (Näätänen & Picton, 1987). Response presence does not verify 2  that the brain has interpreted and perceived the acoustic stimulus (Vanaja, 2016). The information given by CAEPs is limited in that it cannot provide evidence of the brain’s ability to process and integrate complex acoustic signals, such as speech (Näätänen & Picton, 1987).  A clinically observed response absence does not provide definitive evidence of a lack of auditory capacity of the participant. A number of confounding factors could result in a lack of response, such as a high noise floor, or technical difficulties with the recording procedure or stimuli (Lightfoot, 2016; Vanaja, 2016). The CAEP electrophysiological method is effective in determining adult audiometric thresholds to within 10 dB of behavioural thresholds in normal hearing individuals, as well as those with hearing loss (Tomlin, Rance, Graydon, & Tsialios, 2006; Tsui, Wong, & Wong, 2002; Van Maanen & Stapells, 2005). The frequency specificity of CAEPs allows for a high level of accuracy in individual threshold seeking (Coles & Mason, 1984). Despite good correlation between behavioural and objective thresholds, a discrepancy may be observed in some cases (Tsui et al., 2002). Due to the multitude of potential factors influencing CAEP results, further testing is recommended to verify the degree of hearing loss in these cases (Tsui et al., 2002).  CAEPs are currently used in clinic as a method of hearing threshold estimation for individuals who are unable to provide reliable behavioural responses (Hyde, Matsumoto, Alberti, & Li, 1986). Objective measures are of particular importance in the assessment of medicolegal cases such as occupational hearing loss where the prospect of compensation is involved (Boniver, 2002; Dejonckere & Coryn, 2000). There are a number of situations in which the claimant may possess an incentive to exaggerate their thresholds (Hyde et al., 1986). In addition to the investigation of pseudohypacusis, CAEPs provide insight when subjects possess poor task comprehension. These cases may include individuals with low intellectual capabilities, fatigue, 3  or a language barrier (Hyde et al., 1986). The passive paradigm does not require any active participation or overt response from the test subject to measure their physiological auditory capacity (Stapells, 2002). The test procedures present a non-invasive and clinically feasible alternative for situations in which conventional audiometric procedures are not an option (Hyde et al., 1986).  CAEP threshold testing requires expert clinicians who are well trained in recognizing CAEP waveforms that are larger than the background (pre-stimulus) noise. Currently, there are no standards or objective criteria for identifying CAEPs compared to the background noise. One problem with lack of standards in electrophysiological measures of hearing thresholds is in the reliability of CAEP threshold measurements performed, from one clinician to the next. One clinician might be fairly conservative in requiring the pre-stimulus EEG noise level to be low while another clinician might be more liberal. This could lead to different threshold estimations among clinicians. Setting objective EEG noise criteria within the standard clinical practice might help to reduce subjective bias, similar to what is recommended for noise criteria when conducting auditory brainstem response (ABR) measures of hearing thresholds in infants (Hatton, Hyde, & Stapells, 2012). Thus, the aim of my thesis is to investigate and recommend a minimum EEG noise criterion for CAEP testing of hearing in adults.   1.2 Subject-related Factors A number of subject factors can influence the amplitude, latency, and morphology of the N1-P2 response. The waveform is sensitive to endogenous processes and factors separate from the properties of the stimulus. CAEP waveforms undergo a number of complex maturational changes throughout childhood and adolescence (Wunderlich & Cone-Wesson, 2006). The location and orientation of 4  the N1 neural generators undergo change as the brain develops (Bruneau, Roux, Guerin, Barthelemy & Lelord, 1997). N1-P2 latency decreases and the morphology of the response becomes more prominent and less variable with age, becoming fully mature by late adolescence (Wunderlich & Cone-Wesson, 2006). No significant age-related changes in the N1 or P2 amplitude or latency have been identified when young adult CAEPs were compared to those of elderly participants (Polich, 1997). The participant’s degree of attention to the acoustic stimulus also an impact on N1-P2 amplitude. Response amplitude of N1 and P2 were found to be much larger when participants are instructed to count the stimuli compared to reading (Picton & Hillyard, 1974; Mast & Watson, 1986). A less variable response is exhibited in participants who are given reading material, compared to those asked to sit quietly and ignore the stimuli (Keating & Ruhm, 1971). It was noted that individuals who intentionally exaggerate behavioural thresholds often have clear CAEP responses, as their stimulus-related anxiety results in increased attention (Hyde et al., 1986). Recordings during sleep showed a large N2 amplitude, with significant variability depending on stages of sleep (Osterhammel, Davis, Wier, & Hirsh, 1973).  Sleep state introduces variables that are difficult to control, therefore it is often avoided clinically. Furthermore, CAEP thresholds are shown to be elevated above behavioural threshold levels in drowsy and sleeping participants (Mendel, Hosick, Windman, Davis, Hirsh, & Dinges, 1975). Particularly at low intensities, response detection becomes difficult and the probability of clinician error increases (Rapin, Schimmel, & Cohen, 1972).  An individual’s degree of hearing loss has been shown to affect both the latency and amplitude of CAEP responses (Oates, Kurtzberg, & Stapells, 2002). Although changes were noted in the N1 waveform, more significant effects were observed in the later event-related 5  potentials (ERPs) (Oates et al., 2002). Studies have reported increased detectability of threshold level CAEP responses in participants with sensorineural hearing loss (Alanazi, Nicholson, Atcherson, & Martin, 2016; Bardy, Sjahalam-King, Van Dun, & Dillon, 2016; Durante, Wieselberg, Roque, Carvalho, Pucci, Gudayol, & Almeida, 2016; Morita, Naito, Nagamine, Fujiki, Shibasaki, & Ito, 2003). Compared to normal hearing individuals, the hearing loss group exhibited larger N1 amplitudes for stimuli at low presentation levels (Morita et al., 2003). They also showed steeper CAEP amplitude growth curves, possibly due to differences in loudness recruitment in individuals with sensorineural hearing loss (Morita et al., 2003). This enhancement of activation was also noted in participants with idiopathic sudden sensorineural hearing loss (Morita, Hiraumi, Fujiki, Naito, Nagamine, Fukyuyama, & Ito, 2007). The study documented the recovery of the auditory cortex over a period of days (Morita et al., 2007). Subjects with simulated mild conductive hearing loss exhibited a reduction in speech-evoked CAEP response amplitude at soft presentation levels (Munro, Purdy, Ahmed, Begum, & Dillon, 2011). Although CAEP responses for high intensity speech stimuli remained visible, detectability of responses close to hearing thresholds was suboptimal (Munro et al., 2011). 1.3 Stimulus-related Factors  The N1-P2 response is affected by the intensity, frequency, spectral complexity, duration, and inter-stimulus interval of the acoustic signal. These exogenous variables are dependent on the presence and characteristics of the stimuli (Stapells, 2002). The adjustment of stimulus parameters can be used as a strategy to maximize the response amplitude, thus increasing detection sensitivity.   Response amplitude and latency are influenced by the intensity of the stimulus. As the intensity of an auditory stimulus decreases, the N1 response amplitude decreases and the latency 6  increases (Davis & Zerlin, 1966; Onishi & Davis, 1968; Picton, Woods, Baribeau-Braun, & Healey, 1977; Picton, Goodman, & Bryce, 1970). Near threshold, responses become more difficult to identify due to the small N1-P2 amplitude (Davis & Zerlin, 1966; Näätänen & Picton, 1987). The N1-P2 waveform is usually recognizable to within 10 dB of behavioural thresholds, making it an excellent clinical tool for providing evidence of biological hearing capacity (Tsui et al., 2002; Van Maanen & Stapells, 2005).  The N1-P2 response amplitude has been shown to decrease in size with increasing frequency of the stimulus tone (Antinoro, Skinner, & Jones, 1969; Jacobson, Lombardi, Gibbens, Ahmad, & Newman, 1992). This effect was more significant for tonal stimuli with a frequency above 1000 Hz and at higher intensities (Antinoro et al., 1969). With increasing frequency, a decrease in the latency of N1 was noted, particularly between 250 and 1000 Hz (Jacobson et al., 1992).   The N1 response is evoked by a change in the properties of the stimulus from the established baseline. A longer stimulus rise time results in increased N1 latency; this effect was more pronounced at lower stimulus intensities (Kodera, Hink, Yamada, & Suzuki, 1979; Onishi & Davis, 1968). Response amplitude remains relatively constant with a rise time up to 50 ms (Onishi & Davis, 1968). Beyond this point, a longer rise time results in a smaller N1-P2 amplitude (Onishi & Davis, 1968). Stimulus duration also has an impact on response properties. N1 amplitude increases with stimulus plateau duration (Alain, Woods, & Covarrubias, 1997). Response amplitude growth begins to decrease with a stimulus duration greater than 72 ms (Alain et al., 1997). A longer stimulus duration allows for increased frequency specificity by minimizing the spread of excitation in the cochlea. 7  A longer inter-stimulus interval (ISI) also increases the amplitude of the response, with the largest N1-P2 amplitude obtained with an interval of 10 seconds (Davis et al., 1966). A study by Picton et al. found the most efficient ISI to obtain an optimal signal to noise ratio (SNR) within the minimum amount of time to be between 1 and 2 seconds (Picton et al., 1977). The rate of recovery between stimuli is dependent not only on the most recent stimulus, but on the average number of responses per minute (Davis et al., 1966). The summation of multiple responses results in an increased recovery time (Davis et al., 1966). The impact of ISI on the N1 amplitude is less for lower intensity stimuli (Näätänen & Picton, 1987). 1.4 Response Detection 1.4.1 Background Noise The EEG activity recorded during CAEP testing includes the brain’s response to the stimulus, as well as a variety of unrelated potentials. A neural response following the acoustic stimulus is referred to as the signal, and all other potentials make up the residual noise (RN) level (Picton, Linden, Hamel & Maru, 1983). Sources of noise include a variety of exogenous and endogenous aspects of sound and processing (Stapells, 2002). It can present as an electrical artifact, or be related to a non-auditory physiological event such as myogenic noise or the varying of neural potentials due to attentive state (Stapells, 2002).  In order for a response to be properly identified, the N1-P2 waveform must be clearly visible above the level of noise. In CAEP recordings, a high signal to noise ratio (SNR) and low RN value are considered to be favourable (Picton, 2010). Visual response recognition becomes substantially more difficult with low intensity stimuli, as the N1-P2 amplitude is much smaller and the SNR is lower (Hoppe, Weiss, Stewart, & Eysholdt, 2001). In situations with a high noise floor and poor SNR, the background noise may obscure a response. 8  1.4.2 Averaging Waveform averaging is a powerful and commonly used method of noise reduction to overcome a poor SNR. The purpose is to distinguish an auditory response to the stimulus from noise originating from other bodily functions unrelated to the signal (Picton et al., 1983). This is accomplished through repeated presentation of the stimulus, as several waveform replications are combined into an average waveform (Picton et al., 1983).  Background noise unrelated to auditory function presents randomly and has no phase relation to the signal (Picton et al., 1983). Thus noise potentials do not replicate with the same timing in each trial and are effectively cancelled out as a result of averaging (Picton et al., 1983). The auditory response to the stimulus remains mostly constant and time-locked to the stimulus onset for each trial. Thus, the signal becomes more prominent in the average waveform as the amplitude of noise is reduced (Picton et al., 1983). Recording a greater number of replications decreases the standard error present in the average waveform (Schimmel, Rapin, & Cohen, 1974). The SNR and RN remain useful tools for assessing the resulting averaged waveform (Picton et al., 1983). The number of sweeps required to reach an adequate SNR at threshold intensities is dependent on the N1-P2 amplitude and the level of noise present.  1.4.3. Methods of Response Classification  The current standard method of CAEP response classification is clinician judgment. This introduces a level of subjectivity to results of an otherwise objective test (Hyde et al., 1986). Reliable visual interpretation of CAEP waveforms requires adequate training and clinical experience (Hyde et al., 1986).  The British Columbia Early Hearing Program (BCEHP) has established guidelines surrounding the noise floor criteria in auditory-brainstem response (ABR) testing (Hatton et al., 9  2012). In order to classify an average waveform as containing no response, it must be visually flat and have an RN value of less than or equal to 0.08 µV (Hatton et al., 2012). An average waveform with a present response should be replicable and have an SNR of 1.0 or greater (Hatton et al., 2012).  No such criteria exist to guide clinicians in the classification of CAEP thresholds. A study by Picton et al. discussed the need for objective guidelines to overcome the challenge of differentiating a response from residual background noise (1977). It was proposed that a control recording should be taken and kept as a template for reference during the assessment (Picton et al., 1977). In determining a positive response, the waveform must match the same amplitude and latency, with the appropriate morphology for the given stimulus intensity level (Picton et al., 1997). Other studies employed the use of empirically derived standards for amplitude and latency to assist with threshold response detection (Coles & Mason, 1984; Lightfoot & Kennedy, 2006). A study by Yeung and Wong relied on the agreement of two experienced independent examiners in determining the lowest intensity at which the N1 waveform was visible (2007). The techniques used to ensure accuracy of results vary between CAEP testing protocols.  Several studies have explored the possibility of objective response classification using automated machine scoring and statistical analysis (Golding Dillon, Seymour, & Carter, 2009). Feature extraction involving the template of a normal hearing adult CAEP response has been used successfully to determine the presence of a response (Hoth, 1993). Concern was raised regarding the possibility of overlooking responses with components that lie outside the defined template (Golding et al., 2009). Amplitude, latency, and morphology characteristics can vary significantly between and within participants (Golding et al., 2009). Another detection method involved the use of a discrete wavelet transform for feature extraction, followed by statistical 10  analysis (Hoppe et al., 2001). This method performed with higher accuracy compared to the judgment of expert reviewers at threshold intensities (Hoppe et al., 2001). A study by Ross et al. demonstrated the effectiveness of a computerized phase-coherence approach in the detection of significant waveforms (Ross, Lütkenhöner, Pantev, & Hoke, 1999). Recent studies have applied Hotelling’s T2 statistical test to determine response presence (Alanazi et al., 2016; Durante et al., 2016; Golding et al., 2009; Van Dun, Dillon, & Seeto, 2015). These approaches have been shown to perform at or above the level of human expertise under most conditions. It is clear that the use of threshold searching algorithms for response detection represents a promising direction for future research. However, an established maximum RN criterion is recommended before such devices are introduced in clinical practice (Van Dun et al., 2015). As the majority of clinicians continue to use visual judgement to determine response presence, the need for standardized criteria arises. Such criteria can assist clinicians with knowing when to stop recording because an acceptable background noise has been met, making the assessment process more efficient and making judgements more reliable among clinicians. If RN criteria are met, it can be confidently stated that a response could not be obscured by noise in the waveform. 1.5 Goals of Study This study seeks to observe the effect of EEG background noise on a rater’s ability to distinguish between present or absent N1-P2 responses at threshold level intensities. The minimum N1-P2 response amplitude at threshold levels was estimated to be 0.64 μV based on previous research study results at threshold and supra-threshold intensities (Angel, 2016; Picton et al., 1970; Picton et al., 1977; Picton, 2010; Zerlin & Davis, 1966). The RN level must be low enough for a clinician to confidently state that no response is present below threshold intensities. 11  The maximum RN value that would allow for the detection of a small N1-P2 response should therefore be 0.32 μV. A noise floor above this level may possibly obscure a small response. Thus, the goal of my thesis is to estimate a noise criterion based on recorded and simulated CAEP averages with various levels of RN. The results of this study will inform future research, possibly leading to the implementation of new RN guidelines within the standard CAEP test procedure. The inclusion of an RN criterion may provide increased validity and reliability in clinical practice.                12  Chapter 2: Methods 2.1 Participants This study included 12 normal hearing individuals (9 female; 3 male) with an age range of 21 to 30 years of age. All participants reported no hearing-related concerns and had normal hearing thresholds (≤ 20 dB HL between 250 to 4000 Hz). The study was reviewed and approved by the Behavioural Research Ethics Board of the University of British Columbia. Study procedures were reviewed with each participant prior to obtaining informed consent. Datasets of five participants were excluded from the study due to excessive levels of noise present in the recordings. CAEPs were unclear and not interpretable even after the maximum number of allowable replications had been reached. Thus, the remaining datasets were used for further analysis.  2.2 General Study Procedures 2.2.1 Audiometric Assessment Otoscopy was conducted on each study participant, followed by tympanometry using a conventional 226 Hz probe tone to ensure normal middle ear pressure and function. A pure-tone air conduction hearing screening was conducted using ER-3A insert earphones in a double walled sound booth at the UBC School of Audiology and Speech Sciences Electrophysiology Lab. Participants with behavioural hearing thresholds below 20 dB HL from 250 Hz through 4000 Hz bilaterally were considered to be normal and therefore met the inclusion criteria of this study.  13  2.2.2 Electrophysiological Measures Electrophysiological measures were conducted in a double-walled sound booth with the participant sitting comfortably in a reclining chair. During CAEP testing, participants were instructed to relax with their eyes open, and given the option to read. The total session time ranged from approximately 1.5 to 2.5 hours. Short breaks were provided as needed.  Stimulus Parameters. Pure tone air-conduction stimuli were presented through an ER-3A insert earphone at 500 Hz and 2000 Hz to the right ear, while the left ear was occluded with a foam earplug. The total duration of each brief tone was 60 ms, including a 10 ms rise, 40 ms plateau, and 10 ms fall. The stimulus rate was set to 0.87/s. Stimuli were generated by the Intelligent Hearing System (IHS) SMART-EP program, and delivered to the participant through ER-3A insert earphones. Stimuli were calibrated in dB nHL (normal Hearing Level) using a sound-level meter. Calibrations for 0 dB eHL (estimated Hearing Level) were 31 dB and 28 dB for the 500 Hz and 2000 Hz tones, respectively. CAEP Recording Parameters. EEG was collected using the IHS SMART-EP hardware and software. Electrodes were placed on the vertex (CZ), left mastoid (M1), right mastoid (M2), and the forehead ground. Ipsilateral (M2-CZ) and contralateral (M1-CZ) channels were recorded. EEG signals were amplified 100,000 times and filtered through a band pass filter of 1 Hz to 15 Hz. The recording window was set to a total duration of 940 ms, with a 250 ms pre-stimulus for a baseline measurement (i.e., -250 to 690 ms). The sampling rate was 714 Hz. Artifact rejection limits were set to ±51 μV in order to eliminate the effects of non-auditory physiological ERPs such as myogenic artifacts. A set of 20 recording sweeps was averaged for each replication at each stimulus intensity at a rate of 0.87 sweeps/s (or 1.11 s/sweep). A total of 10 replications of 14  20 sweeps were collected at threshold and 5 dB below the hearing threshold. Other intensities were often required to determine threshold as described below.  CAEP Recording Procedure. A modified Hughston-Westlake ascending bracketing procedure was used for CAEP testing at 500 Hz and 2000 Hz in the right ear. The assessment began with a stimulus intensity of 0 dB nHL. Following a positive response, stimulus intensity was decreased in 10 dB steps. Following an observed “no response,” stimulus intensity was increased in 5 dB steps to determine hearing thresholds. CAEP thresholds were found at intensities whereby a CAEP was judged by the researcher to be “present,” and whereby the researcher judged CAEP responses to be “absent” at an intensity that was 5 dB lower.  Judgement of response absence or presence at each intensity was performed by a trained audiology student and verified retrospectively by an experienced researcher. Recordings (200 sweeps) for the threshold level intensity whereby CAEP responses were determined to be present (referred to as “Present Recordings”), as well as the 200 sweep sub-threshold recordings (referred to as “Absent Recordings”), were organized in a bank of 280 waveforms to be used for presenting to the Experienced Raters as described below.    2.3 CAEP Rating Procedure Two independent experts volunteered to view the bank of 280 waveforms (7 participants x 2 channels x 2 conditions [present/absent] x 2 frequencies x 5 RN levels) and judge whether N1-P2 responses were present or absent in the waveforms presented to them from the bank. Waveforms were displayed in the following format using a computer monitor (Figure 2.1). For each judgment, two replications were displayed as thin overlapping grey lines. The average waveform appeared below the replications as a single thick black line. The waveforms presented consisted of a varying number of sweeps: 40 (2 reps of 20 sweeps), 80 (2 reps of 40 sweeps), 120 15  (2 reps of 60 sweeps), 160 (2 reps of 80 sweeps), and 200 (2 reps of 100 sweeps) sweeps. A thin vertical line at 0 ms represented the onset of the stimulus.  Raters were blinded from information regarding the frequency, intensity, SNR, and RN level for the waveforms presented. Threshold and sub-threshold level recordings with varying RN levels were presented randomly interleaved, to ensure that judges would be unable to predict patterns between the waveforms. For each judgment the raters were instructed to indicate whether an N1-P2 response was absent or present. Raters were not given the option to reject waveforms containing high residual noise levels. Following the CAEP judgments, raters reported the strategy that they had applied in determining whether or not a response was present in each waveform.  Figure 2.1: Format for Rating CAEP Thresholds  Waveform display format presented to raters for the purpose of judging response presence/absence for both recorded and simulated CAEPs. The thin grey lines represent two replications at a given frequency and intensity, each containing half of the sweeps contained within the average waveform. The thick grey line represents the average waveform comprised of the two replications. The thin vertical line at 0 ms indicated the onset of the auditory stimulus. The y-axis displayed amplitude in microVolts, and the x-axis represented time in milliseconds.  16  2.4 Simulated CAEP Data A bank of simulated waveforms (37 N1-P2 amplitudes by 8 RN values) was created based on N1-P2 amplitude values from data collected from a previous study (Angel, 2016). This dataset includes gap-detection CAEP recordings at threshold intensities for 47 individuals. The N1-P2 response amplitudes for the CAEP gap threshold data ranged from 0.6403 μV to 2.9825 μV, and were used to generate the simulated CAEP waveforms for the current study. Simulated “signal” waveforms were generated for 40 sweeps with durations of -250 to 690 ms, with a sampling rate of 714 Hz. Each signal waveform was created from three Gaussian windows of various durations and amplitudes representing the P1, N1, and P2 responses of the CAEPs. Durations randomly ranged from 35-65 ms, 50-75ms, and 60-120 ms for the P1, N1, and P2 responses, respectively. The centre of the Gaussian windows (i.e., P1, N1, P2 peaks) randomly ranged from 45-55, 90-120, and 160-230 ms. Amplitudes for the Gaussian windows randomly ranged from between 20-80, 50-80, and 50-100 percent of the waveform amplitude set to -1 to 1, respectively for the P1, N1, and P2 waves. The Gaussian windows were then summed together to form the P1-N1-P2 waveform for each sweeps. The sweeps were then multiplied by a gain factor such that the averaged waveform across the 40 sweeps had a threshold-like N1-P2 amplitude that selected from a bank of 37 uniformly space amplitudes ranging from 0.6403 μV to 2.9825 μV. This “signal only” dataset of 40 sweeps was added to a noise dataset of 40 sweeps, as described below. A pink-noise generator (power value randomly ranging from 1.2 to 1.5) was used to simulate noise waveforms with target pre-stimulus noise levels of 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, and 1.6 μV. These noise levels were created by randomly sampling and gaining 40 sweeps of pink noise such that the standard deviation of the averaged noise sweeps in the pre-stimulus 17  interval and the post-stimulus RN value resulted in the target noise level. These 40 sweeps were then split into 2 replications of 20 sweeps and were used as the “Absent” simulated response waveforms.   2.5 Simulated CAEP Data Rating Procedure Three independent experts volunteered to view the bank of simulated CAEP data and judge whether N1-P2 responses were present or absent. Each rater judged 592 waveforms (37 datasets x 2 conditions [present/absent] x 8 RN levels) presented in a random order. The waveforms were displayed following the exact format and procedure outlined above in section 2.3 for judgment of recorded CAEP responses.   2.6 Statistical Analysis  Pre-stimulus noise and post-stimulus RN levels were measured for each rater under both recorded and simulated conditions. The pre-stimulus noise level was defined as the standard deviation of the average waveform within the interval of -250 to 0 ms. The post-stimulus RN was calculated as the standard deviation of the plus-minus reference between the two replications (Picton, 2010, p. 157-158). The post-stimulus RN was measured between 0 and 250 ms.  A true positive judgement was defined for a “Present” N1-P2 waveform when the rater selected the “Yes” responses (aka, Hit). A false negative judgement was defined for a “Present” N1-P2 waveform when the rater selected the “No” responses (aka, Miss). A true negative judgement was defined for an “Absent” N1-P2 waveform when the rater selected “Yes” responses (aka, Correct Rejection). A false positive judgement was defined for an “Absent” N1-P2 waveform when the rater selected “Yes” responses (aka, False Alarm). The rate of these judgements types were then used to calculate the sensitivity index (d’) and response bias for each rater.   18  The rate of true-positive and false-positive judgments were also binned relative to pre-stimulus noise level. A sigmoidal curve was fitted to these binned data to estimate the sensitivity and specificity for each rater. The pre-stimulus noise criterion was estimated from the fitted curve wherein the N1-P2 responses were correctly identified with 95% sensitivity. The post-stimulus RN values were also measured. Because these values linearly changed relative to the pre-stimulus noise levels, the post-stimulus RN criterion was estimated from the pre-stimulus noise level at 95% sensitivity. This was carried out for the results of each rater’s judgements of recorded and simulated data.  An inter-rater reliability analysis was performed in order to ascertain the consistency between the raters. First, the percent agreement among the raters was calculated for each condition. The expected agreement that could occur due to random chance was also calculated. Taking the previous calculations into account, inter-rater reliability was determined using the Cohen’s Kappa statistical method (Fleiss & Cohen, 1973). The above calculations were performed for the raters’ responses to the recorded data, as well as for the simulated data.      19  Chapter 3: Results 3.1 Recorded and Simulated CAEP Results CAEP results for seven participants that were included in the recorded CAEP bank of waveforms are plotted in Figure 3.1. The CAEPs are plotted in order of increasing standard deviation ratio (SDR), which is a ratio of the standard deviation of the average waveform to the standard deviation of the plus-minus reference (Picton, 2010, p. 158). Simulated CAEPs for 37 participants were included in the second bank of CAEP waveforms and are plotted together in Figure 3.2. Threshold (“Present”) and sub-threshold (“Absent”) level recordings were presented with a range of RN values. Pink noise was applied at 8 different levels to each CAEP in order to generate waveforms with the following pre-stimulus RN values: 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, and 1.6 μV. As with the recorded data, CAEPs in the simulated waveform bank were randomized and presented to three raters.             20   Figure 3.1: Recorded CAEP Results Recorded CAEP waveforms for seven participants are plotted in order of increasing standard deviation ratio (SDR) from left to right. Each average waveform contains 200 sweeps. Ipsilateral and contralateral channel recordings of stimuli at 500 Hz and 2000 Hz were included. The green lines represent CAEPs with present N1-P2 responses and black lines represent CAEPs with absent N1-P2 responses. Each CAEP was presented to the raters a total of five times, consisting of a varying numbers of sweeps (40, 80, 120, 160, and 200 sweeps). The y-axis displays the ERP amplitude in microVolts, and the x-axis represented time in milliseconds.       21   Figure 3.2: Simulated CAEP Results Simulated CAEP waveforms for 37 participants are plotted for each RN level that was judged by the three raters. The pre-stimulus RN level increases from left to right: 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4, 1.6 μV. The green lines represent CAEPs with present N1-P2 responses and black lines represent CAEPs with absent N1-P2 responses. The y-axis displays the ERP amplitude in microVolts, and the x-axis represented time in milliseconds.       22  3.2 Rater Judgments for Recorded and Simulated CAEPs Waveform judgment results for Rater 1, Rater 2, and Rater 3 are seen in Figure 3.3, Figure 3.4, and Figure 3.5, respectively. Figure 3.3 and 3.4 include both recorded and simulated CAEP results. Graphs A and B display the true-positive and false-positive rates as a function of pre-stimulus noise level for recorded and simulated data, respectively. Graphs C and D are plots of the post-stimulus RN value and the pre-stimulus noise level for the recorded and simulated data, respectively.  Figure 3.5 contains only simulated data, because Rater 3 was the experimenter who collected the original recorded data and thus would be biased to performing judgments on recorded data. For this figure, Graph A displays the true positive and false positive rates as a function of pre-stimulus noise level. Graph B is the plot of the post-stimulus RN value by the pre-stimulus noise level. Overall, the rate of true positives increased as a function of decreasing pre-stimulus noise level. The rate of false positives remained relatively low across all judgments. Results from the simulated CAEP judgments exhibit the expected sigmoidal curve typically associated with signal detection theory. Recorded results exhibited a different pattern, with a higher true positive rate in CAEPs with increased noise level. The point at which raters were able to correctly detect a “present” response with 95% sensitivity was indicated on each graph by a pink asterisk. As predicted, this value consistently corresponded with a low pre-stimulus noise level.  There is a clear correlation between the post-stimulus RN value and the pre-stimulus noise level for the recorded and simulated data. The noise level measured in the pre-stimulus interval increases as a function of post-stimulus RN level. 23   Figure 3.3: Rater 1 Judgments for Recorded and Simulated CAEPs The above graphs relate Rater 1’s true positive and false positive scores to the pre-stimulus and post-stimulus noise level. Graphs A and B display the true positive and false positive rates as a function of pre-stimulus noise level for recorded and simulated data, respectively. Graphs C and D represent the correlation between the post-stimulus RN value and the pre-stimulus noise level for the recorded and simulated data, respectively. The green circles represent true positive responses, indicating that the rater correctly judged an N1-P2 response to be present within the CAEP. The red X symbols represent false positive responses, indicating that the rater incorrectly judged an N1-P2 response to be present. The pink asterisk indicates the point at which the rater can correctly detect a present response with 95% sensitivity.  24   Figure 3.4: Rater 2 Judgments for Recorded and Simulated CAEPs  The above graphs relate Rater 2’s true positive and false positive scores to the pre-stimulus and post-stimulus noise level. Graphs A and B display the true positive and false positive rates as a function of pre-stimulus noise level for recorded and simulated data, respectively. Graphs C and D represent the correlation between the post-stimulus RN value and the pre-stimulus noise level for the recorded and simulated data, respectively. The green circles represent true positive responses, indicating that the rater correctly judged an N1-P2 response to be present within the CAEP. The red X symbols represent false positive responses, indicating that the rater incorrectly judged an N1-P2 response to be present. The pink asterisk indicates the point at which the rater can correctly detect a present response with 95% sensitivity.  25   Figure 3.5: Rater 3 Judgments for Simulated CAEPs  The above graphs relate Rater 3’s true positive and false positive scores to the pre-stimulus and post-stimulus noise level. Graph A displays the true positive and false positive rates as a function of pre-stimulus noise level for the simulated data. Graph B represents the correlation between the post-stimulus RN value and the pre-stimulus noise level for the recorded and simulated data, respectively. The green circles represent true positive responses, indicating that the rater correctly judged an N1-P2 response to be present within the CAEP. The red X symbols represent false positive responses, indicating that the rater incorrectly judged an N1-P2 response to be present. The pink asterisk indicates the point at which the rater can correctly detect a present response with 95% sensitivity.   26  Following the waveform judgments, raters described the strategies used in determining the presence or absence of an N1-P2 response. All three raters compared the post-stimulus standard deviation of the possible response to the standard deviation of noise within the pre-stimulus region. In addition, Rater 1 assessed the consistency of the noise level between the pre-stimulus and post-stimulus regions.  Amplitude, latency, and morphology of the potential response within the post-stimulus interval were assessed by all raters. Raters 2 and 3 reported evaluating the replicability between the two individual replications before evaluating the average waveform.  3.3 Inter-Rater Reliability The percent agreement, percent chance, and kappa statistic were calculated to assess inter-rater reliability among the judges. Calculations included two raters for the recorded dataset and three raters for simulated dataset, shown in Table 3.1. Inter-rater reliability (Cohen’s kappa) was remarkably similar for recorded and simulated conditions. Cohen’s kappa for both recorded data and simulated data indicates a moderate level of agreement between raters. Although the simulated CAEPs yielded a higher percent agreement, this was counterbalanced by an increased level of expected chance agreement.   Recorded CAEPs Simulated CAEPs Percent Agreement 76.43% 82.88% Percent Chance 49.86% 64.23% Cohen’s kappa  0.5298 0.5215  Table 3.1: Inter-rater Reliability for all CAEP Judgments  27  3.4 Sensitivity Index and Response Bias   The sensitivity index (d’) and response bias were calculated for the recorded and simulated datasets, for each rater (Table 3.2). For both recorded and simulated conditions, all raters demonstrated good discriminability. The results of Rater 2 exhibited the highest sensitivity index, indicating an excellent ability to distinguish between CAEPs with absent and present responses. Response bias was consistently higher for the simulated CAEP judgments, compared to the recorded CAEPs. This suggests that raters employed a more conservative set of criteria when judging the presence of N1-P2 responses in the simulated waveforms. Larger positive response biases refer to more conservative judgements (less desire to make false positives with the trade-off of having fewer true positive), whereas less positive (or negative) response biases refer to more liberal judgements (more likely to make false positives with the trade-off of having more true positives). No judgment was performed by Rater 3 for the recorded dataset.  Sensitivity Index (d’) Response Bias  Recorded Simulated Recorded Simulated Rater 1 1.559 1.1787 - 0.0121 0.5131 Rater 2 1.9167 1.4156 0.7601 1.3404 Rater 3 -  1.0352 - 0.9267  Table 3.2: Sensitivity Index and Response Bias for Recorded and Simulated CAEPs  3.5 Pre-stimulus and Post-stimulus Noise Levels The noise level at which each rater was able identify present N1-P2 responses at 95% sensitivity was estimated for recorded and simulated CAEPs. The pre-stimulus and post-stimulus noise levels of the average waveforms wherein these judgments occurred are presented in Table 28  3.3. Based on the recorded CAEP data for Rater 2, it was not possible to estimate the point at which 95% sensitivity occurred, likely because of the rater’s higher response bias. Rater 3 did not perform judgements for the recorded waveforms.   Post-stimulus RN Level Pre-stimulus Noise Level  Recorded Simulated Recorded Simulated Rater 1 0.215 μV 0.208 μV 0.105 μV 0.181 μV Rater 2 -  0.145 μV -  0.111 μV Rater 3 -  0.166 μV -  0.105 μV  Table 3.3: Noise Levels for True Positive Judgements at 95% Sensitivity               29  Chapter 4: Discussion  In the interpretation of CAEP results, a clinician must determine the absence or presence of an N1-P2 response. The minimum amplitude of an N1-P2 response at threshold intensities was estimated to be 0.6471 μV based on previous research study results at threshold and supra-threshold intensities (Angel, 2016; Picton et al., 1970; Picton et al., 1977; Picton, 2010; Zerlin & Davis, 1966). A high noise floor may obscure a small N1-P2 response. As such, the RN level must be sufficiently low in order for a clinician to confidently state that “no response” is present below the threshold level. Because RN decreases as a function of the number of sweeps, clinicians can record more sweeps to include in the average to meet a specific noise criterion. The goal of this thesis is to estimate an RN criterion based on recorded and simulated CAEP averages that vary in the level of noise. This information could provide guidance for clinicians in deciding that the noise is sufficiently low to be able to confidently state “no response” is present during the evaluation of CAEP waveforms at threshold intensities. 4.1 CAEPs and Noise Level Measures  For both recorded and simulated data, the rate of true positives increased as a function of decreasing pre-stimulus noise level. The rate of false positives remained relatively low across all noise levels. In a clinical context, false positive errors could result in severe consequences. In addition, false negative errors often trigger further investigation. Medicolegal cases also demand a high level of accuracy. Understandably, clinicians have strict response judgment criteria, only rating responses as being present if they feel highly confident. For CAEPs with high levels of noise, raters consistently demonstrated a tendency of erring on the side of caution that responses were not present. This lead to a low pre-stimulus noise levels at the point in which judges had 95% sensitivity. 30  4.1.1 Recorded CAEPs   The bank of recorded CAEP waveforms contained a relatively small sample size. There was an overall lack of N1-P2 response variability, causing many of the waveforms to look similar to one another across the averages of different number of sweeps (i.e., 40, 80, 120, 160, and 200). This is partially due to the fact that the same signal was captured in most of the averages. In addition, ipsilateral and contralateral recordings were presented as separate waveforms to increase the number of waveforms to judge but the variability in signal amplitude and morphology was limited to seven participants’ data. When recorded CAEP results were plotted on a graph, the pattern differed from the expected sigmoidal curve that is associated with signal detection theory. The recorded dataset may not exhibit this trend due to the lack of N1-P2 amplitude variation and waveform morphology. It is possible that raters could have visualized a response template, and applied it to waveforms with a higher RN. The true positive rate could have been artificially elevated due to the raters’ ability to recognize waveforms within the small sample size. This results in a poor level of face validity among recorded CAEP results. This was one reason why the simulated-data condition was included in this study. The simulated data had 296 (37*8) different signal morphologies and 37 different N1-P2 amplitudes; providing greater variability that is likely more representative of larger samples of CAEP threshold data. However, future work investigating clinical judgements on real recorded CAEP datasets from a larger sample size is recommended. 4.1.2 Simulated CAEPs The simulated CAEP dataset included a larger sample size than the recorded CAEP dataset. The bank of waveforms included a wide variety of N1-P2 amplitudes in a linear equal distribution. The level of diversity seen in the simulated data was more representative of the 31  general population. CAEP thresholds were validated against behavioural thresholds in order to definitively determine whether or not a response was present in each CAEP. The noise floor level was manipulated with great precision by applying pink noise to each waveform to match specific pre-stimulus and RN noise targets. The simulated CAEP judgments followed a sigmoidal curve, which is typically associated with signal detection theory. These results thus have a significantly higher face validity, compared to the results of the recorded CAEPs.  4.1.3 Pre-stimulus and Post-stimulus Noise Level Measures The majority of EEG recording systems calculate the RN level based on the standard deviation of the noise (plus-minus references) within the post-stimulus interval. In this study, the post-stimulus RN value was measured between 0 and 250 ms, as well as the pre-stimulus noise level. The reason for this is because clinicians mostly make their response judgements by comparing the post-stimulus averaged waveform to the pre-stimulus interval. Raters reported observing the pre-stimulus region as an effective strategy to assess the replicability and noise level of a given waveform. Unfortunately, most clinical EEG systems that can be used to conduct CAEP testing do not calculate a pre-stimulus noise level. If the pre-stimulus noise level were made available to clinicians, this would serve as a meaningful measure of the noise floor. This is why post-stimulus RN values were also measured.  4.1.4 Selection of an RN Level Criterion The minimum N1-P2 response amplitude within the simulated data was set to be 0.64 μV, based on previous study (Angel, 2016). This value is consistent with estimates of the minimum N1-P2 amplitude from previous studies (Picton et al., 1970; Picton et al., 1977; Picton, 2010; Zerlin & Davis, 1966). The predicted maximum RN value that would allow for the detection of a small N1-P2 response that is at least two time larger that the noise, should 32  therefore be 0.32 μV. In this study, the minimum pre-stimulus and post-stimulus noise levels determined by the three raters with 95% sensitivity ranged from 0.145 to 0.215 μV, which are below 0.32 μV. This might be due to the fact that CAEPs thresholds for clinical evaluations typically include waveforms from higher intensities when judging response presence or absence. The additional information may improve response detection. For the purpose of this study, the raters were blinded to CAEPs at higher intensities and thus the RN noise criterion values are lower than that predicted from the literature.   Compared to the recorded CAEP dataset, the simulated results were more generalizable. The simulated data included a larger sample size and featured a broad array of N1-P2 response amplitudes, ranging from 0.64 μV to 2.98 μV. Furthermore, the CAEP thresholds for the simulated results were validated against behavioural thresholds. The results of Rater 2’s judgments yielded the largest sensitivity index, indicating a higher level of accuracy, and had more conservative response bias. Because of this, Rater 2’s judgments were considered to be the better of all raters for the simulated CAEPs, and Rater 2’s noise criterion at 95% sensitivity were chosen as the better estimates. Thus, the recommended pre-stimulus noise level is 0.111 μV and the recommended post-stimulus RN is 0.145 μV. When the noise floor is at or below this level, a clinician can confidently state that there is no response obscured by background noise. If the noise floor is above the recommended RN criterion, a small N1-P2 response could still be present because it is obscured by background EEG noise.  4.2 Inter-rater Reliability The kappa for both recorded data and simulated data indicates a moderate level of agreement between raters. There was a relatively high level of agreement between raters, particularly for the simulated CAEP data. The percentage of expected chance was likely elevated 33  by the large number of instances where raters scored responses as being absent. Due to the high level of noise in many of the CAEPs, raters were unable to confidently detect the presence of an N1-P2 response. Clinicians may exercise caution in an effort to minimize false positive judgments. As such, the increased level of chance likely resulted in a decreased overall kappa score for inter-rater reliability. 4.3 Caveats High levels of noise presented a particular challenge during the recorded CAEP datasets in this study. Recorded results for five participants were excluded from the data analysis due to excessive background EEG noise present in the CAEPs. For these cases, it was unclear whether or not a response was present, even upon reaching the maximum allowable 200 sweeps at each intensity. This resulted in an extended recording time for participants with high noise levels. Thus, it may not be feasible for clinicians to reduce noise sufficiently in order to reach the minimum RN criterion for all patients. The small sample size for the recorded CAEPs also represents a limitation of this study as discussed previously. For both recorded and simulated CAEPs, raters were blinded to the frequency, intensity, RN level, and SNR of each waveform. In addition, the CAEPs were displayed individually with no recordings above or below the intensity of the single waveform presented. Under realistic circumstances, a clinician has this contextual information at their disposal; this can aid in informing the decision of whether or not an N1-P2 response is present. A forced judgement of “response present” or “response absent” is not completely generalizable, as clinicians often have the opportunity to either record an additional replication, or classify the CAEP as “could not evaluate.” For this reason, the minimum RN criterion derived from this study should be considered to be a conservative estimate.  34  4.4 Clinical Implications and Directions for Future Research The results of this study clearly demonstrate the importance of achieving a low RN level in recordings. The recommended pre-stimulus noise level of 0.111 μV or post-stimulus RN level of 0.145 μV can serve as a helpful conservative guideline for clinical practice. As visual judgment is currently the standard means of determining the presence or absence of a response, a specific criterion could assist clinicians in interpreting results. Due to the small N1-P2 amplitude at threshold intensities, an RN criterion would be particularly beneficial. A noise floor criterion has been successfully implemented within the BCEHP protocol for ABR assessment (Hatton et al., 2012). If such a criterion were introduced for CAEP interpretation, this could increase reliability among clinicians’ judgments. Prior to the implementation of an RN criterion for CAEPs, a thorough cost-benefit analysis must take place. The implementation of a strict RN criterion may result in longer test time; however, this could yield more accurate results. The consequences of false positive and false negative judgments must be evaluated within a clinical context, in order to determine an acceptable degree of error. False positive errors could result in an underestimation of hearing loss, which may cause insufficient amplification or untreated hearing loss. False negative errors may lead to overestimation of hearing loss and unnecessary amplification.  Within a medicolegal context, CAEP threshold testing is performed to confirm behavioural thresholds. If CAEP thresholds are lower than behavioural thresholds, then pseudohypacusis is suspected. By using a RN criterion, as recommend in this thesis, more accurate threshold results may be obtained; thereby leading to better detection of pseudohypacusis. Furthermore, if the EEG noise is large in some cases because clinicians are not 35  using an RN criterion, then false negative errors may occur in pseudohypacusis cases and lead to unnecessary compensation. Future research should include a larger sample size in order to capture greater variability across participants, as seen in the simulated CAEP results. Further investigation is needed for different populations, such as children and individuals with hearing loss. Studies should also investigate the use of an RN criterion when raters are allowed to simultaneously view waveforms at multiple intensities. It would be highly informative to evaluate the effect of an RN criterion implemented in a clinical context, when compared to the current standard of visual judgments of single intensity waveforms.   4.5 Conclusions The present study investigated the effects of the RN on a clinicians’ ability to judge the presence or absence of N1-P2 responses in CAEPs. The results were relatively consistent between the simulated and recorded datasets. The results of this study provide evidence supporting the use of an RN criterion to assist in determining whether or not a response is present. The use of an RN criterion may improve the accuracy by which clinicians can interpret CAEP results. Furthermore, this could increase reliability among the judgments of multiple clinicians. Future research should include a larger sample size and investigate the noise floor in a more clinical context. Further studies may lead to the implementation of new guidelines surrounding RN criteria in CAEP recording procedures. Such guidelines may provide more validity and reliability in clinical practice.    36  Bibliography  Alain, C., Woods, D. L., & Covarrubias, D. (1997). Activation of duration-sensitive auditory cortical fields in humans. Electroencephalography and Clinical Neurophysiology/Evoked Potentials Section, 104(6), 531-539. Alanazi, A. A., Nicholson, N., Atcherson, S. R., & Martin, P. (2016). The HEARLab cortical tone evaluation (cte) protocol: a clinical feasibility study. Journal of Hearing Science, 6(3), 54-62. Antinoro, F., Skinner, P. H., & Jones, J. J. (1969). Relation between sound intensity and amplitude of the AER at different stimulus frequencies. The Journal of the Acoustical Society of America, 46(6B), 1433-1436. Angel, R. (2016). Cortical Auditory Evoked Potentials to Estimate Gap-Detection Thresholds in Adults. cIRcle: UBC’s Digital Repository: Electronic Theses and Dissertations (ETDs) 2008+. Bardy, F., Sjahalam-King, J., Van Dun, B., & Dillon, H. (2016). Cortical auditory-evoked potentials in response to multitone stimuli in hearing-impaired adults. Journal of the American Academy of Audiology, 27(5), 406-415. Bruneau, N., Roux, S., Guerin, P., Barthelemy, C., & Lelord, G. (1997). Temporal prominence of auditory evoked potentials (N1 wave) in 4‐8‐year‐old children. Psychophysiology, 34(1), 32-38. Coles, R. R. A., & Mason, S. M. (1984). The results of cortical electric response audiometry in medico-legal investigations. British journal of audiology, 18(2), 71-78. Davis, P. A. (1939). Effects of acoustic stimuli on the waking human brain. Journal of neurophysiology, 2(6), 494-499. Davis, H., Mast, T., Yoshie, N., & Zerlin, S. (1966). The slow response of the human cortex to auditory stimuli: recovery process. Electroencephalography and clinical neurophysiology, 21(2), 105-113. Davis, H., & Zerlin, S. (1966). Acoustic relations of the human vertex potential. The Journal of the Acoustical Society of America, 39(1), 109-116. Dejonckere, P. H., & Coryn, C. P. (2000). A Comparison Between Middle Latency Responses and Late Auditory Evoked Potentials for Approximating Frequency-Specific Hearing Levels in Medicolegal Patients with Occupational Hearing Loss. International Tinnitus Journal, 6(2), 175-181. Durante, A. S., Wieselberg, M. B., Roque, N., Carvalho, S., Pucci, B., Gudayol, N., & Almeida, K. D. (2017). Assessment of hearing threshold in adults with hearing loss using an automated system of cortical auditory evoked potential detection. Brazilian journal of otorhinolaryngology, 83(2), 147-154. 37  Fleiss, J. L., & Cohen, J. (1973). The equivalence of weighted kappa and the intraclass correlation coefficient as measures of reliability. Educational and psychological measurement, 33(3), 613-619. Golding, M., Dillon, H., Seymour, J., & Carter, L. (2009). The detection of adult cortical auditory evoked potentials (CAEPs) using an automated statistic and visual detection. International Journal of Audiology, 48(12). 833-842. Hatton, J., Hyde, M., & Stapells, D. (2012). BC Early Hearing Program. Audiology Assessment Protocol. Version, 4. Hoppe, U., Weiss, S., Stewart, R. W., & Eysholdt, U. (2001). An automatic sequential recognition method for cortical auditory evoked potentials. IEEE Transactions on Biomedical Engineering, 48(2), 154-164. Hoth, S. (1993). Computer-aided hearing threshold determination from cortical auditory evoked potentials. Scandinavian audiology, 22(3), 165-177. Hyde, M., Matsumoto, N., Alberti, P., & Li, Y. L. (1986). Auditory evoked potentials in audiometric assessment of compensation and medicolegal patients. Annals of Otology, Rhinology & Laryngology, 95(5), 514-519.  Jacobson, G. P., Lombardi, D. M., Gibbens, N. D., Ahmad, B. K., & Newman, C. W. (1992). The effects of stimulus frequency and recording site on the amplitude and latency of multichannel cortical auditory evoked potential (CAEP) component N1. Ear and Hearing, 13(5), 300-306. Jones, S. J., Longe, O., & Pato, M. V. (1998). Auditory evoked potentials to abrupt pitch and timbre change of complex tones: electrophysiological evidence of streaming? Electroencephalography and Clinical Neurophysiology/Evoked Potentials Section, 108(2), 131-142. Keating, L. W., & Ruhm, H. B. (1971). Within average variability of the acoustically evoked response. Journal of speech and hearing research, 14(1), 179-188. Kodera, K., Hink, R. F., Yamada, O., & Suzuki, J. I. (1979). Effects of rise time on simultaneously recorded auditory-evoked potentials from the early, middle and late ranges. Audiology, 18(5), 395-402. Lightfoot, G. (2016). Summary of the N1-P2 cortical auditory evoked potential to estimate the auditory threshold in adults. Seminars in hearing 37(1), 1-8. Thieme Medical Publishers. Lightfoot, G., & Kennedy, V. (2006). Cortical electric response audiometry hearing threshold estimation: accuracy, speed, and the effects of stimulus presentation features. Ear and hearing, 27(5), 443-456. Mast, T. E., & Watson, C. S. (1968). Attention and auditory evoked responses to low-detectability signals. Perception & Psychophysics, 4(4), 237-240. 38  Mendel, M. I., Hosick, E. C., Windman, T. R., Davis, H., Hirsh, S. K., & Dinges, D. F. (1975). Audiometric comparison of the middle and late components of the adult auditory evoked potentials awake and asleep. Electroencephalography and clinical neurophysiology, 38(1), 27-33. Morita, T., Hiraumi, H., Fujiki, N., Naito, Y., Nagamine, T., Fukuyama, H., & Ito, J. (2007). A recovery from enhancement of activation in auditory cortex of patients with idiopathic sudden sensorineural hearing loss. Neuroscience research, 58(1), 6-11. Morita, T., Naito, Y., Nagamine, T., Fujiki, N., Shibasaki, H., & Ito, J. (2003). Enhanced activation of the auditory cortex in patients with inner-ear hearing impairment: a magnetoencephalographic study. Clinical neurophysiology, 114(5), 851-859. Munro, K. J., Purdy, S. C., Ahmed, S., Begum, R., & Dillon, H. (2011). Obligatory cortical auditory evoked potential waveform detection and differentiation using a commercially available clinical system: HEARLab™. Ear and Hearing, 32(6), 782-786. Näätänen, R., Gaillard, A. W., & Mäntysalo, S. (1978). Early selective-attention effect on evoked potential reinterpreted. Acta psychologica, 42(4), 313-329. Näätänen, R., & Picton, T. (1987). The N1 wave of the human electric and magnetic response to sound: a review and an analysis of the component structure. Psychophysiology, 24(4), 375-425. Oates, P. A., Kurtzberg, D., & Stapells, D. R. (2002). Effects of sensorineural hearing loss on cortical event-related potential and behavioral measures of speech-sound processing. Ear and hearing, 23(5), 399-415. Osterhammel, P. A., Davis, H., Wier, C. C., & Hirsh, S. K. (1973). Adult auditory evoked vertex potentials in sleep. Audiology, 12(2), 116-128. Picton, T. W. (2010). Human auditory evoked potentials. Plural Publishing. Picton, T. W., Goodman, W. S., & Bryce, D. P. (1970). Amplitude of evoked responses to tones of high intensity. Acta oto-laryngologica, 70(2), 77-82. Picton, T. W., & Hillyard, S. A. (1974). Human auditory evoked potentials. II: Effects of attention. Electroencephalography and clinical neurophysiology, 36, 191-200. Picton, T. W., Linden, R. D., Hamel, G., & Maru, J. T. (1983, November). Aspects of averaging. In Seminars in Hearing (Vol. 4, No. 04, pp. 327-340). Copyright© 1983 by Thieme Medical Publishers, Inc. Picton, T. W., Woods, D. L., Baribeau-Braun, J., & Healey, T. M. (1977). Evoked potential audiometry. J Otolaryngol, 6(2), 90-119. Polich, J. (1997). EEG and ERP assessment of normal aging. Electroencephalography and Clinical Neurophysiology/Evoked Potentials Section, 104(3), 244-256. 39  Rapin, I., Schimmel, H., & Cohen, M. M. (1972). Reliability in detecting the auditory evoked response (AER) for audiometry in sleeping subjects. Electroencephalography and clinical neurophysiology, 32(5), 521-528. Ross, B., Lütkenhöner, B., Pantev, C., & Hoke, M. (1999). Frequency-specific threshold determination with the CERAgram method: basic principle and retrospective evaluation of data. Audiology and Neurotology, 4(1), 12-27. Schimmel, H., Rapin, I., & Cohen, M. M. (1974). Improving evoked response audiometry with special reference to the use of machine scoring. Audiology, 13(1), 33-65. Stapells, D. R. (2002). Cortical event-related potentials to auditory stimuli. Handbook of clinical audiology, 5, 378-406. Tomlin, D., Rance, G., Graydon, K., & Tsialios, I. (2006). A comparison of 40 Hz auditory steady-state response (ASSR) and cortical auditory evoked potential (CAEP) thresholds in awake adult subjects. International Journal of Audiology, 45(10), 580-588. Tsui, B., Wong, L. L., & Wong, E. C. (2002). Accuracy of cortical evoked response audiometry in the identification of non-organic hearing loss. International journal of audiology, 41(6), 330-333. Vanaja, C. S. (2016). Cortical Auditory Evoked Potentials: Evidence for use in Clinical Practice. Journal of the All India Institute of Speech & Hearing, 35. Van Dun, B., Dillon, H., & Seeto, M. (2015). Estimating hearing thresholds in hearing-impaired adults through objective detection of cortical auditory evoked potentials. Journal of the American Academy of Audiology, 26(4), 370-383. Van Maanen, A., & Stapells, D. R. (2005). Comparison of multiple auditory steady-state responses (80 versus 40 Hz) and slow cortical potentials for threshold estimation in hearing-impaired adults. International Journal of Audiology, 44(11), 613-624. Wunderlich, J. L., & Cone-Wesson, B. K. (2006). Maturation of CAEP in infants and children: a review. Hearing research, 212(1-2), 212-223. Yeung, K. N., & Wong, L. L. (2007). Prediction of hearing thresholds: Comparison of cortical evoked response audiometry and auditory steady state response audiometry techniques. International Journal of Audiology, 46(1), 17-25. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0384598/manifest

Comment

Related Items