Derivation of input-output functions from distortion-product otoacoustic emission level maps

Zelle, Dennis; Bader, Katharina; Dierkes, L; Gummer, Anthony W.; Dalhoff, Ernst

doi:10.1121/10.0001142

Cited by 11 publications

(11 citation statements)

References 65 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results suggest that in many cases, the hybrid models can make full use of the advantages of DL for feature extraction and traditional ML algorithms for regression decision to further optimize the predictive power of SFOAEs. (Ellison & Keefe 2005) and DPOAE results (Gorga et al 2003;Johnson et al 2007;Dalhoff et al 2013;Zelle et al 2017bZelle et al , 2020. The SFOAE study of Ellison and Keefe (2005) correlated the hearing threshold with SFOAE S/N at the audiometric frequency, whereas the DPOAE studies used the approach originally proposed by Boege and Janssen (2002) to predict hearing thresholds based on an EDPT extracted from the linear regression of I/O functions of DPOAE pressure amplitude versus stimulus level.…”

Section: Sfoae Test Performance Based On DLmentioning

confidence: 99%

“…The SFOAE study of Ellison and Keefe (2005) correlated the hearing threshold with SFOAE S/N at the audiometric frequency, whereas the DPOAE studies used the approach originally proposed by Boege and Janssen (2002) to predict hearing thresholds based on an EDPT extracted from the linear regression of I/O functions of DPOAE pressure amplitude versus stimulus level. An EDPT was accepted as valid only when at least three points on the I/O function had S/N ≥6 or 10 dB and the criteria r 2 ≥ 0.8 or 0.49, s ≥ 0.2 or Zelle et al (2020). The SEs for this study were lower than those for the DPOAE studies of Gorga et al and Johnson et al, but higher in some cases than those for the single-source DPOAE studies (Dalhoff et al;Zelle et al 2017bZelle et al , 2020, which focused on the hearing thresholds below 30 dB HL or a subpopulation of normal-hearing ears.…”

Section: Sfoae Test Performance Based On DLmentioning

confidence: 99%

“…An EDPT was accepted as valid only when at least three points on the I/O function had S/N ≥6 or 10 dB and the criteria r 2 ≥ 0.8 or 0.49, s ≥ 0.2 or Zelle et al (2020). The SEs for this study were lower than those for the DPOAE studies of Gorga et al and Johnson et al, but higher in some cases than those for the single-source DPOAE studies (Dalhoff et al;Zelle et al 2017bZelle et al , 2020, which focused on the hearing thresholds below 30 dB HL or a subpopulation of normal-hearing ears.…”

Section: Sfoae Test Performance Based On DLmentioning

confidence: 99%

“…Attempts have therefore been made to use OAEs, especially distortion-product OAEs (DPOAEs) (Gorga et al 1993a, b, 1997; Kim et al 1996; Stover et al 1996; Dorn et al 1999; Norton et al 2000; Go et al 2019) and transient-evoked OAEs (Prieve et al 1993; Hurley & Musiek 1994; Lichtenstein & Stapells 1996; Hussain et al 1998; Norton et al 2000; Mertes & Goodman 2013), to predict hearing status. Going beyond the determination of hearing status, many studies attempted to estimate hearing thresholds based on the extrapolated distortion-product thresholds (EDPTs) from the linear regression of input-output (I/O) functions of DPOAE pressure amplitude versus primary level (Boege & Janssen 2002; Gorga et al 2003; Johnson et al 2007; Dalhoff et al 2013; Zelle et al 2017a, b, 2020).…”

Section: Introductionmentioning

confidence: 99%

“…Performance comparison in threshold prediction between the proposed SFOAE-based DL models and the SFOAE study ofEllison and Keefe (2005), the DPOAE studies ofGorga et al (2003) andJohnson et al (2007), the single-source DPOAE studies ofZelle et al (2017bZelle et al ( , 2020 andDalhoff et al (2013) …”

mentioning

confidence: 99%

See 4 more Smart Citations

Deep Learning Models for Predicting Hearing Thresholds Based on Swept-Tone Stimulus-Frequency Otoacoustic Emissions

Liu,

Gong

2023

Ear &Amp; Hearing

View full text Add to dashboard Cite

Objectives: This study aims to develop deep learning (DL) models for the quantitative prediction of hearing thresholds based on stimulus-frequency otoacoustic emissions (SFOAEs) evoked by swept tones. Design: A total of 174 ears with normal hearing and 388 ears with sensorineural hearing loss were studied. SFOAEs in the 0.3 to 4.3 kHz frequency range were recorded using linearly swept tones at a rate of 2 Hz/msec, with stimulus level changing from 40 to 60 dB SPL in 10 dB steps. Four DL models were used to predict hearing thresholds at octave frequencies from 0.5 to 4 kHz. The models—a conventional convolutional neural network (CNN), a hybrid CNN-k-nearest neighbor (KNN), a hybrid CNN-support vector machine (SVM), and a hybrid CNN-random forest (RF)—were individually built for each frequency. The input to the DL models was the measured raw SFOAE amplitude spectra and their corresponding signal to noise ratio spectra. All DL models shared a CNN-based feature self-extractor. They differed in that the conventional CNN utilized a fully connected layer to make the final regression decision, whereas the hybrid CNN-KNN, CNN-SVM, and CNN-RF models were designed by replacing the last fully connected layer of CNN model with a traditional machine learning (ML) regressor, that is, KNN, SVM, and RF, respectively. The model performance was evaluated using mean absolute error and SE averaged over 20 repetitions of 5 × 5 fold nested cross-validation. The performance of the proposed DL models was compared with two types of traditional ML models. Results: The proposed SFOAE-based DL models resulted in an optimal mean absolute error of 5.98, 5.22, 5.51, and 6.06 dB at 0.5, 1, 2, and 4 kHz, respectively, superior to that obtained by the traditional ML models. The produced SEs were 8.55, 7.27, 7.58, and 7.95 dB at 0.5, 1, 2, and 4 kHz, respectively. All the DL models outperformed any of the traditional ML models. Conclusions: The proposed swept-tone SFOAE-based DL models were capable of quantitatively predicting hearing thresholds with satisfactory performance. With DL techniques, the underlying relationship between SFOAEs and hearing thresholds at disparate frequencies was explored and captured, potentially improving the diagnostic value of SFOAEs.

show abstract

Section: Sfoae Test Performance Based On DLmentioning

confidence: 99%

Section: Sfoae Test Performance Based On DLmentioning

confidence: 99%

Section: Sfoae Test Performance Based On DLmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Deep Learning Models for Predicting Hearing Thresholds Based on Swept-Tone Stimulus-Frequency Otoacoustic Emissions

Liu,

Gong

2023

Ear &Amp; Hearing

View full text Add to dashboard Cite

show abstract

Gepulste DPOAE in Verlaufsmessungen

Bader,

Zelle,

Gummer

et al. 2024

HNO

View full text Add to dashboard Cite

Zusammenfassung Hintergrund Bisher gibt es keinen Konsens darüber, wie Ototoxizität in Verlaufsmessungen standardisiert zu erfassen ist. Für die Diagnostik von Schädigungen des cochleären Verstärkers sind Messverfahren notwendig, die eine möglichst hohe Test-Retest-Zuverlässigkeit und eine hohe Aussagekraft hinsichtlich persistierender Schädigungen aufweisen. Hörschwellenschätzungen auf der Grundlage von Kurzpuls-DPOAE-Pegelkarten („estimated distortion-product thresholds“, LEDPT) berücksichtigen individuell optimale DPOAE-Anregungspegel und erlauben eine zuverlässige quantitative Schätzung des cochleär bedingten Hörverlusts. Methodik Hörschwellen wurden mithilfe von LEDPT objektiv geschätzt und mit einer modifizierten Békésy-Tracking-Audiometrie (LTA) subjektiv erfasst. Die Messungen wurden siebenmal innerhalb von drei Monaten bei 14 Frequenzen (f2 = 1–14 kHz) in 20 Ohren (PTA4 (0,5–4 kHz) < 20 dB HL) durchgeführt. Die Rekonstruktion des DPOAE-Wachstumsverhaltens in Abhängigkeit von den Anregungspegeln L1,L2 erfolgte auf der Grundlage von 21 DPOAE-Amplituden und ermöglichte mithilfe einer numerischen Anpassung einer nichtlinearen mathematischen Funktion die Berechnung eines LEDPT für jede Anregungsfrequenz. Für die gleichzeitige kombinierte Betrachtung wurden Verteilungen der Hörschwellen (LTA, LEDPT), der DPOAE-Pegel (LDP) und Kombinationen davon ermittelt. Ergebnisse Einzeln betrachtet wiesen LTA und LEDPT jeweils eine Test-Retest-Zuverlässigkeit mit einem Median der absoluten Differenzen (AD) von 3,2 dB bzw. 3,3 dB auf, der sich durch Anwendung eines kombinierten Analyseparadigmas aus LEDPT, LDP und LTA auf 2,0 dB signifikant reduzieren ließ. Schlussfolgerung Es ist zu erwarten, dass ein auf einer Kombination von LEDPT, überschwelligen LDP, und feinstrukturreduzierter LTA basierendes Analyseparadigma eine höhere Güte (Sensitivität und Spezifität) des Tests erzielt, um pathologische oder auch regenerative Veränderungen der äußeren Haarsinneszellen zuverlässig zu detektieren.

show abstract

Pulsed DPOAEs in serial measurements

Bader,

Zelle,

Gummer

et al. 2024

HNO

View full text Add to dashboard Cite

Background To date, there is no consensus on how to standardize the assessment of ototoxicity in serial measurements. For the diagnosis of damage to the cochlear amplifier, measurement methods are required that have the highest possible test-retest reliability and validity for detecting persistent damage. Estimated distortion-product thresholds (LEDPT) based on short-pulse distortion-product otoacoustic emission (DPOAE) level maps use individually optimal DPOAE stimulus levels and allow reliable quantitative estimation of cochlea-related hearing loss. Materials and methods Hearing thresholds were estimated objectively using LEDPT and subjectively using modified Békésy tracking audiometry (LTA). Recordings were performed seven times within three months at 14 frequencies (f2 = 1–14 kHz) in 20 ears (PTA4(0.5–4 kHz) < 20 dB HL). Reconstruction of the DPOAE growth behavior as a function of the stimulus levels L1, L2 was performed on the basis of 21 DPOAE amplitudes. A numerical fit of a nonlinear mathematical function to the three-dimensional DPOAE growth function yielded LEDPT for each stimulus frequency. For the combined analysis, probability distributions of hearing thresholds (LTA, LEDPT), DPOAE levels (LDP), and combinations thereof were determined. Results LTA and LEDPT each exhibited a test-retest reliability with a median of absolute differences (AD) of 3.2 dB and 3.3 dB, respectively. Combining LEDPT, LDP, and LTA into a single parameter yielded a significantly smaller median AD of 2.0 dB. Conclusion It is expected that an analysis paradigm based on a combination of LEDPT, suprathreshold LDP, and fine-structure-reduced LTA would achieve higher test performance (sensitivity and specificity), allowing reliable detection of pathological or regenerative changes in the outer hair cells.

show abstract

Derivation of input-output functions from distortion-product otoacoustic emission level maps

Cited by 11 publications

References 65 publications

Deep Learning Models for Predicting Hearing Thresholds Based on Swept-Tone Stimulus-Frequency Otoacoustic Emissions

Deep Learning Models for Predicting Hearing Thresholds Based on Swept-Tone Stimulus-Frequency Otoacoustic Emissions

Gepulste DPOAE in Verlaufsmessungen

Pulsed DPOAEs in serial measurements

Contact Info

Product

Resources

About