A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions
This article describes a Bayesian active-learning procedure for estimating the edge frequency, f e , of a dead region, that is, a region in the cochlea with no or very few functioning inner hair cells or neurons. The method is based on the psychophysical tuning curve (PTC) but estimates the shape of...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SAGE Publishing
2018-07-01
|
| Series: | Trends in Hearing |
| Online Access: | https://doi.org/10.1177/2331216518788215 |
| _version_ | 1818497909290696704 |
|---|---|
| author | Josef Schlittenlacher Richard E. Turner Brian C. J. Moore |
| author_facet | Josef Schlittenlacher Richard E. Turner Brian C. J. Moore |
| author_sort | Josef Schlittenlacher |
| collection | DOAJ |
| description | This article describes a Bayesian active-learning procedure for estimating the edge frequency, f e , of a dead region, that is, a region in the cochlea with no or very few functioning inner hair cells or neurons. The method is based on the psychophysical tuning curve (PTC) but estimates the shape of the PTC from the parameters of a hearing model, namely f e , and degree of outer hair cell loss. It chooses the masker frequency and level for each trial to be highly informative about the model parameters in the context of previous data. The procedure was tested using 14 ears from eight subjects previously diagnosed with high-frequency dead regions. The estimates of f e agreed well with estimates obtained using “Fast PTCs” or more extensive measurements from an earlier study. On average, 33 trials were needed for the estimate of f e to fall and stay within 0.3 Cams of the final “true” value on the equivalent rectangular bandwidth-number scale. The time needed to obtain a reliable estimate was 5 to 8 min. This is comparable to the time required for Fast PTCs and short enough to be used when fitting a hearing aid. Compared with Fast PTCs, the new method has the advantage of using yes-no judgments rather than continuous Békésy tracking. This allows the slope of a subject’s psychometric function and thus the reliability of his or her responses to be estimated, which in turn allows the test duration to be adjusted so as to achieve a given accuracy. |
| first_indexed | 2024-12-10T18:51:20Z |
| format | Article |
| id | doaj.art-f360a2569e4c4973be08b3b0c7e95b6c |
| institution | Directory Open Access Journal |
| issn | 2331-2165 |
| language | English |
| last_indexed | 2024-12-10T18:51:20Z |
| publishDate | 2018-07-01 |
| publisher | SAGE Publishing |
| record_format | Article |
| series | Trends in Hearing |
| spelling | doaj.art-f360a2569e4c4973be08b3b0c7e95b6c2022-12-22T01:37:18ZengSAGE PublishingTrends in Hearing2331-21652018-07-012210.1177/2331216518788215A Hearing-Model-Based Active-Learning Test for the Determination of Dead RegionsJosef Schlittenlacher0Richard E. Turner1Brian C. J. Moore2Department of Experimental Psychology, University of Cambridge, UKDepartment of Engineering, University of Cambridge, UKDepartment of Experimental Psychology, University of Cambridge, UKThis article describes a Bayesian active-learning procedure for estimating the edge frequency, f e , of a dead region, that is, a region in the cochlea with no or very few functioning inner hair cells or neurons. The method is based on the psychophysical tuning curve (PTC) but estimates the shape of the PTC from the parameters of a hearing model, namely f e , and degree of outer hair cell loss. It chooses the masker frequency and level for each trial to be highly informative about the model parameters in the context of previous data. The procedure was tested using 14 ears from eight subjects previously diagnosed with high-frequency dead regions. The estimates of f e agreed well with estimates obtained using “Fast PTCs” or more extensive measurements from an earlier study. On average, 33 trials were needed for the estimate of f e to fall and stay within 0.3 Cams of the final “true” value on the equivalent rectangular bandwidth-number scale. The time needed to obtain a reliable estimate was 5 to 8 min. This is comparable to the time required for Fast PTCs and short enough to be used when fitting a hearing aid. Compared with Fast PTCs, the new method has the advantage of using yes-no judgments rather than continuous Békésy tracking. This allows the slope of a subject’s psychometric function and thus the reliability of his or her responses to be estimated, which in turn allows the test duration to be adjusted so as to achieve a given accuracy.https://doi.org/10.1177/2331216518788215 |
| spellingShingle | Josef Schlittenlacher Richard E. Turner Brian C. J. Moore A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions Trends in Hearing |
| title | A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions |
| title_full | A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions |
| title_fullStr | A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions |
| title_full_unstemmed | A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions |
| title_short | A Hearing-Model-Based Active-Learning Test for the Determination of Dead Regions |
| title_sort | hearing model based active learning test for the determination of dead regions |
| url | https://doi.org/10.1177/2331216518788215 |
| work_keys_str_mv | AT josefschlittenlacher ahearingmodelbasedactivelearningtestforthedeterminationofdeadregions AT richardeturner ahearingmodelbasedactivelearningtestforthedeterminationofdeadregions AT briancjmoore ahearingmodelbasedactivelearningtestforthedeterminationofdeadregions AT josefschlittenlacher hearingmodelbasedactivelearningtestforthedeterminationofdeadregions AT richardeturner hearingmodelbasedactivelearningtestforthedeterminationofdeadregions AT briancjmoore hearingmodelbasedactivelearningtestforthedeterminationofdeadregions |