Objective Measurement of Distortion Within the Auditory System
July 2018
Kimberly Jenkins, AuD, CCC-A
The profession of audiology is evolving, both the manner in which we evaluate our patients and the rehabilitative strategies to help listeners with perceived hearing difficulties. One of the most common complaints of individuals seeking audiologic care is a decline in the ability to comprehend speech in the presence of competing noise. Clinical diagnostic tests include the pure-tone audiogram (the current gold standard for hearing assessment) and speech testing (in quiet and in noise). There is growing frustration within the audiology community as more and more audiologists find that results from standard diagnostic tests do not always explain patient complaints or predict benefit from rehabilitative treatment, particularly when individuals seeking care have normal results on hearing assessments. Even the plethora of available speech-in-noise tasks has yet to offer consistent prediction of people's communication ability in everyday listening environments.
This prompted numerous audiologic research projects, including in-depth physiologic analyses, improved diagnostic measures, and more focused rehabilitative strategies. Plomp (1978, 1986) and Middelweerd, Festen, and Plomp (1990) anticipated this problem, including the limited benefit of hearing aids in noise and reverberation, which led them to develop their speech reception threshold (SRT) model of hearing handicap. The model assumes that hearing loss for speech can be accounted for by the sum of two simple factors: a reduction in the level of both speech and noise (attenuation factor) as measured primarily by an audiogram, and a distortion factor represented by a decrease in the speech-to-noise ratio (SNR loss). This latter component has been discussed in terms of broadened auditory filters, abnormal loudness growth, and reduced dynamic range; however, attempts to measure individual auditory processing distortion or to understand its acoustic and physiological properties has been difficult. In 2009, Kujawa and Liberman published a report of (a) irreparable damage to the ribbon synapses between inner hair cells and the auditory nerve and (b) delayed degeneration of spiral ganglion cells in mice following temporary threshold shifts due to noise exposure. Prior to this discovery, the conventional thought was that if the pure-tone audiogram yielded normal hearing thresholds (i.e., audibility is no longer a factor), then the periphery of the auditory system was undamaged and functioning normally—therefore, any complaints reported by the patient were assumed to be due to suprathreshold signal distortion, central auditory processing deficits, or cognitive issues. However, with newfound knowledge of the physiologic repercussions of noise, it is becoming apparent that the pure-tone audiogram and speech testing do not provide us with adequate information to effectively perform our jobs. Improved diagnostic tests are therefore critical to better meet the needs of the clinician's patient population.
One of the major drawbacks to the current diagnostic test battery is its failure to pinpoint subtle deficits of signal processing within the auditory system. As is suggested by the early modeling efforts by Plomp and colleagues (1978, 1986) and by findings from Kujawa and Liberman (2009), noise-related damage may be present but may go undetected on a pure-tone audiogram. This damage could result in degradation of an acoustic signal (speech, most notably), which may negatively influence communication ability in the presence of noise but would not necessarily be significant enough to be detected with standard speech testing. Better identification of the pathology causing patients' hearing issues or how the acoustic input is processed by each individual may therefore prove advantageous for the clinician.
But how can we achieve this? Increasingly, researchers and clinicians are turning to auditory electrophysiologic [EEG] measures for answers. These objective measures utilize an array of acoustic signals to evoke responses along the auditory pathway—from the periphery through the auditory cortex—and may provide additional information regarding the functioning of the auditory system not uncovered by the current diagnostic test battery.
Certain EEG measures are already used in audiology clinics, although not consistently. Electrocochleography (known as ECOG or EChoG) measures potentials that arise from the activity of outer hair cells, the synapse between inner hair cells and the auditory nerve (eighth nerve), and the distal portion of the eighth nerve. In theory, this would be an ideal measure to collect on individuals with normal- to near-normal auditory thresholds who nevertheless report having hearing difficulties in noise to better understand the functioning of the periphery—particularly to detect the presence of synaptopathy as discussed by Kujawa and Liberman (2009). In practice, this measure is extremely difficult to obtain because the source of the potential is at the level of the cochlea, which is encased within the temporal bone. Even with the use of tiptrodes (canal electrodes) or tymptrodes (canal electrodes that touch the tympanic membrane), the measurement is variable within an individual and between test sessions. The auditory brainstem response (ABR) is another common measure collected in the clinic (mainly in infants for hearing testing—less so with adults) that provides information about the function of the distal portion of the auditory nerve as well as the function of the nuclei of the auditory brainstem. As with ECOG, there is variable success in collecting Wave I of the ABR (important for assessing synaptic function), and there is not a one-to-one correlation between the additional waves of the response and the nuclei within the brainstem; thus, detailed information regarding auditory pathway function cannot be obtained. Additionally, the stimuli used to evoke these two potentials are clicks or tone-bursts, neither of which are ecologically valid nor predict performance in real-world communication environments. James Jerger (2014) has recommended the use of later event-related potentials that are evoked using speech and contextual cues (i.e., mismatched negativity, late positive complex). These potentials can, in theory, offer insight into how the signal is being processed and interpreted by the individual. However, these measures are highly variable, both from person to person and at the individual level between sessions. Also, because measures are significantly influenced by state of arousal, the provider must be extremely cautious when interpreting results from these measures. It is therefore imperative to find a measure that (a) has high fidelity within an individual between sessions, (b) is not influenced by patient state of arousal, and (c) can provide information regarding the auditory processing of incoming signals. One potential immediately comes to mind—the frequency following response (FFR)—and, in the future, it may prove to be an invaluable clinical tool for diagnostic and rehabilitative assessment.
The FFR, also known as the envelope following response (EFR) or complex ABR, is a unique measurement of sustained phase-locked encoding to periodic stimuli arising largely from the inferior colliculus that faithfully encodes amplitude fluctuations of input stimuli up to 1500 Hz. This measure largely preserves spectral and temporal components of the incoming acoustic signals, as shown in Figure 1.
Figure 1. Input stimulus "da" and the resulting response from an individual. Low- and mid-frequency amplitude fluctuations are preserved within the response curve.
Using this fact to our advantage, a number of analyses can be performed that provide information about the auditory system's ability to process incoming signals and that may hold the key to more effective diagnostic testing.
At Walter Reed National Military Medical Center (WRNMMC), research is underway within the Audiology and Speech Center to uncover more sensitive diagnostic test measures for military personnel who experience perceived difficulties hearing in noise but have normal-hearing auditory thresholds. One measure being investigated extensively is the FFR to synthetic speech syllables. Responses of 2,000 sweeps each to the syllable "/da/" are currently being collected from military personnel with normal-hearing thresholds. Within this group, we are comparing the data of individuals who report no issues with hearing speech in the presence of background noise and competing speakers and the data of those who report being blast-exposed and having significant issues with hearing speech in the presence of background noise. A number of analyses are underway that may provide insight into why some individuals report significant hearing issues when their diagnostic tests are normal. Because the syllable "/da/" is composed of a region transitioning from the consonant "d" to the vowel "a" (transition region) as well as the steady "a" (steady state region), we can analyze the neural processing of each component as well as the syllable as a whole. When comparing the amplitudes of the response between groups, we are finding that those who report no perceived hearing deficits have more robust responses than those who report problems (see Figure 2). What is more interesting, however, is what is seen when comparing the resting neural activity of the two groups (pre-stimulus response). The group that reports issues has an increased amount of neural activity in times of silence, which effectively reduces the speech-to-noise ratio and may be interfering with their ability to encode incoming speech signals in the presence of background noise.
Figure 2. Average root-mean-square (RMS) amplitude of the response for each region of the syllable "/da/" and the entire response. Also shown is the baseline RMS amplitude obtained when no stimulus is played (pre-stimulus).
When we compare the resting neural activity to the activity when the signal is presented, we also note something remarkable. As seen in Figure 3, individuals who do not report having problems hearing in noise have higher signal-to-noise ratios (SNRs) in their neural responses. This may further indicate that the increased spontaneous neural firing measured in those with perceived hearing deficits is interfering with their ability to encode the incoming speech signal, thus making it more difficult to hear speech in background noise.
Figure 3. Signal-to-noise ratio of the two groups for the three regions of the signal.
As previously mentioned, the preservation of temporal and spectral components of the signal within this response can be used to our advantage during analysis. Because the response mimics the stimulus, we can correlate how accurately the response is encoded by performing stimulus-to-response correlations. Although our data appear to show no significant differences between groups, it is, nevertheless, an important aspect in the diagnostic process. We are still in the midst of data collection, and as more individuals are added to the study, there may prove to be significant differences in the neural processing of the signal. What has shown to be more intriguing is an analysis of internal stability within an individual. Each person's 2,000-sweep response was divided into two responses based on 1,000 randomly selected sweeps, then compared to the remaining 1,000 sweeps. The average of those two were correlated. This was repeated for each individual 100 times, and an average correlation value was obtained. Theoretically, because the participant is receiving the identical stimulus 2,000 times, if his or her auditory system processed the sound in a robust and similar manner each time the stimulus was presented, the cross-correlation values would be high. Lower correlation values could indicate greater instability within the auditory system, which may contribute to issues hearing in noise. As you can see from the data collected (see Figure 4), the group with hearing issues encodes speech with more variability than the control group. What is compelling is that, within the test group, these internal stability measures correlate nicely with scores on a subjective survey of hearing (Speech, Spatial, and Qualities of Hearing Scale – SSQ; Gatehouse & Noble, 2004) (data not shown). Summarizing all of the data, it is clear that individuals who report having hearing deficits do, in fact, process sound differently than those who do not report such problems.
Figure 4. Group comparisons of internal stability measure within individuals.
There is still much more work to be done regarding the utility of this measure, but one thing is clear: Complex ABR measures can provide us with information about how an individual processes sound—information that cannot be obtained from the pure-tone audiogram or from speech testing.
How does this information affect the clinician and future patient care? There is tremendous potential for this measure to direct treatment options for patients. For example, if the clinician determines that an individual has a poor SNR value due to increased internal physiologic noise, then that individual may benefit from a low-gain hearing aid—whereas if the individual's internal stability or stimulus-to-response correlation is poor, auditory training may be a better option. If no deficits are present within the FFR, it may be that cognitive deficits or outside influence (sleep deprivation, stress, etc.) are the problem—in which case, outside referrals should be made as necessary. As the research grows for the utility of these electrophysiological measures, we may find that it could someday be an invaluable tool to the clinician. Our current clinical practices fall short of allowing us to provide effective patient care, and it is time to start thinking outside the box to solve this problem. The use of EEG measures, the FFR in particular, may prove to be such a solution because it has the potential to supply us with a great deal of information regarding how an individual processes sound. Therefore, as we obtain more knowledge about what this measure tells us in terms of auditory system functionality, let us consider looking beyond the audiogram and start to think about the use of EEG for more effective and efficient patient care.
The identification of specific products or scientific instrumentation does not constitute endorsement or implied endorsement on the part of the author, United States Department of Defense, or any component agency. The views expressed in this presentation are those of the author and do not reflect the official policy of the Department of Army/Navy/Air Force, Department of Defense, or U.S. Government.
About the Author
Kimberly Jenkins, AuD, CCC-A, received a doctoral degree in clinical audiology from the University of Maryland, College Park, and a bachelor of science degree in biochemistry from Montana Tech of the University of Montana. Before her career in audiology, Dr. Jenkins worked at the National Institute of Mental Health in a schizophrenia lab performing behavioral studies in mice. She has been a research audiologist at Walter Reed National Military Medical Center since January 2017. Prior to joining the research center at Walter Reed, Dr. Jenkins was a clinician at a private neurotologist practice in Rockville, Maryland. Her current research interests include development of electrophysiologic measures for the assessment of functional hearing and communication deficits in the military population. Dr. Jenkins currently serves as associate investigator on two central auditory processing disorder protocols and one hair cell regeneration research protocol within the National Military Audiology and Speech Pathology Center.
References
Gatehouse, S., & Noble, W. (2004). The speech, spatial and qualities of hearing scale (SSQ). International journal of audiology, 43(2), 85–99.
Jerger, J. (2014). 20Q: Using auditory event related potentials for a better understanding of word recognition. AudiologyOnline, Article 12731. Retrieved from https://www.audiologyonline.com/articles/using-auditory-event-related-potentials-12731.
Kujawa, S. G., & Liberman, C. (2009). Adding insult to injury: Cochlear nerve degeneration after "temporary" noise-induced hearing loss. The Journal of Neuroscience, 29(45), 14077–14085. https://doi.org/10.1523/JNEUROSCI.2845-09.2009.
Middelweerd, M. J., Festen, J. M., & Plomp, R. (1990). Difficulties with speech intelligibility in noise in spite of a normal pure-tone audiogram [Original papers]. Audiology, 29(1), 1–7. https://doi.org/10.3109/00206099009081640.
Plomp, R. (1978). Auditory handicap of hearing impairment and the limited benefit of hearing aids. The Journal of the Acoustical Society of America, 63(2), 533–549.
Plomp, R. (1986). A signal-to-noise ratio model for the speech-reception threshold of the hearing impaired. Journal of Speech, Language, and Hearing Research, 29(2), 146–154.
