Acoustical Society of America
158th Meeting Lay Language Papers
Comodulation Masking Release (CMR) as a diagnostic tool to detect cochlear dead regions in hearing-impaired ears
Riya Singh - singh33@illinois.edu
Dept. of Electrical and Computer Engineering
University of Illinois at Urbana-Champaign
Champaign, IL 61820, USA
Woojae Han
Dept. of Speech and Hearing Science
University of Illinois at Urbana-Champaign
Champaign, IL 61820, USA
Jont Allen
Dept. of Electrical and Computer Engineering and Beckman Institute,
University of Illinois at Urbana-Champaign
Champaign, IL 61820, USA
Popular version of paper 2aPP2
Presented Tuesday morning, October 27, 2009
158th ASA Meeting, San Antonio, TX
Cochlear dead regions are places along the basilar membrane of the cochlea where the inner hair cells (IHCs) are non-functioning due to damaged or missing IHC cilia (which are fine "hair bundles" at the top of the cell that transduce the basilar membrane motions). In addition, the afferent auditory neurons innervating those places may be degenerate. A dead region can be described in terms of the characteristic frequency of the IHC at the specific region on the basilar membrane where it occurs. Speech perception is seriously degraded in these regions of degraded transduction. The important conclusions from a study on cochlear dead regions by Brian Moore in 2001 [1], are:
1. “Dead regions maybe relatively common in people with moderate-to-severe sensorinueral hearing loss.”
2. “Dead regions cannot be reliably diagnosed from the pure tone audiogram.”
3. “Psychophysical Tuning Curves (PTCs) provide a useful way of detecting dead regions and defining their boundaries. However, the determination of PTCs is probably too time-consuming to be used for routine diagnosis of dead regions in clinical practice.”
4. “Amplification of frequencies well inside a high-frequency dead region usually does not improve speech intelligibility, and may sometimes impair it.”
Another recent diagnostic tool is the TEN (Threshold Equalizing Noise) test which involves measuring the threshold for detecting a sinusoidal tone presented in a special background noise called the threshold equalizing noise. However, it has been argued that this test is not accurate [2]. Owing to the limitations of the currently used procedures to detect cochlear dead regions, and the fact that the presence or absence of dead regions can have important implications in fitting of hearing-aids, our research group at UIUC has been looking for
alternative methods to detect dead regions. One such technique we are evaluating uses the Comodulation Masking Release (CMR) paradigm.
The CMR effect (by Hall, Haggards, & Fernandes, 1984)
Consider trying to detect a pure tone (target) in a narrow-band noise (masker). The target detection threshold increases as the noise bandwidth increases, as long as it is within the bandwidth of the auditory filter centered on the tone frequency. If we extend the masker to flanking bands, well outside the critical bandwidth, the target threshold depends on the correlations between the modulation envelopes of the target and flanking band maskers. Surprisingly, it becomes easier to detect the tone in spite of having added more noise when the target and flanking maskers are correlated! Thus these flanking bands having coherent modulation produce masking release (Fig 1. from [4]). This effect can be attributed to the ability of the auditory system to make comparisons of envelope fluctuations across frequency.
Figure 1. A just-detectable 700 Hz pure tone is centered on and masked by an envelope-modulated band of noise. When co-modulated narrow-band flanking maskers are added, the detectability increases (From [4] pp.165])
CMR as a diagnostic tool:
While the CMR effect has been extensively studied by many researchers, the contribution of this work is in trying to use CMR as a diagnostic tool to detect isolated cochlear dead regions in hearing-impaired (HI) individuals. Our hypothesis is that, due to loss of tuning in a dead region, there would be no release in masking if the target tone is in a dead region. The CMR test can be performed at any frequency by choosing suitable flanking band frequencies. The current study includes 19 subjects (33 ears) having mild-to-moderate sensorinueral hearing loss. The procedure requires the HI listener to detect a pure tone target in presence of a narrow masker(on-signal band) with four flanking bands in two conditions : 1) the flanking bands have random amplitude modulations 2) the flanking bands are co modulated so to have the same envelope fluctuation as the masker; at various signal-to-noise conditions (-9 to 9 dB). At each SNR and each condition, N (8 or 12) trials are presented to the listener in a 3 interval forced choice method (chance = 33%). One of these intervals contains the target tone. The listener must identify the interval having the target or can choose “SAME” if all three intervals seem identical. The PTC at 1, 2 and 4 kHz is next measured. Finally, a full rank confusion matrix (CM) is measured using nonsense CV syllables under various noise conditions. Since we know the speech cues for these consonants a-priori [5, 6], consonant loss profiles from the CM experiment allow estimates of the possibility of a dead region.
Results:
Figure 2 shows the PTC results of a normal hearing listener TK-R for 1, 2 and 4 kHz. The results indicate no-dead region at these frequencies as confirmed by CMR results (Figure 3), where there is a normal masking release at these frequencies.
Figures 4 (for PTC) and 5 (for CMR) are the results with hearing-impaired ear VS-R (high frequency sloping hearing loss since past 2 years), again tested at 1, 2 and 4 kHz. The dotted lines in figure 4 indicate average normal hearing data. Both PTC and CMR results indicate a dead region at 2-4 [kHz]. This diagnosis is confirmed by the speech perception experiment wherein, listener VS-R has very low scores for CVs /fa/, /sa/, /Sa/, /ba/; having dominant perceptual cues in the 2-4 [kHz] region[5][6].
Figure 2. PTC results for normal hearing listener TK-R
Figure 3. CMR results for normal hearing listener TK-R
Figure 4. PTC results for hearing impaired listener VS-R
Figure 5. CMR results for hearing impaired listener VS-R
Conclusions:
While it is premature to come to a strong conclusion, results appear consistent with the hypothesis that CMR can be used as a diagnostic tool to detect cochlear dead regions. The number and center frequency of the flanking bands can be conveniently chosen so that they fall into regions where we believe there is no dead region as established by CM testing. For unilateral hearing loss, flanking bands can be put in the contralateral ear. With the current experimental setup, we hope to gain insight on how a hearing-impaired ear detects speech modulations and thus form correlations with the ability to understand speech. This is a work in progress.
References:
1. Moore BCJ. Dead regions in the cochlea: Diagnosis, perceptual consequences, and implications for the fitting of hearing aids. Trends in Amplification. 2001
2. Summers V, Molis MR, Musch H, Walden BE, Surr RK, Cord, MT. Identifying Dead Regions in the Cochlea: Psychophysical Tuning Curves and Tone Detection in Threshold-Equalizing Noise. Ear and Hearing, 2003
3. Moore, Brian C. J., Dead Regions in the Cochlea: Conceptual Foundations, Diagnosis, and Clinical Applications. Ear and Hearing, 2004
4. Christopher J. Plack. Sense of Hearing, 1st Edition (May 2005)
5. Li Feipeng, Anjali Menon, and Jont B. Allen, (2009) Perceptual cues in natural speech for 6 stop consonants, J. Acoust. Soc. Am., Resubmission 10/1/09; Revisions requested 7/13/09; submitted: 3/1/09
6. Li Feipeng, Anjali Menon, and Jont B. Allen, (2009) A methodology to study perceptual cues of 8 fricative consonants in natural speech, J. Acoust. Soc. Am., in preparation