Probing cochlear compressive nonlinearity using signal-to-noise ratio–optimized distortion product otoacoustic emission input/output functions

Cochlear compression, a nonlinear response critical for encoding sound dynamics, can be non-invasively assessed using distortion product otoacoustic emission (DPOAE) input/output (i/o) functions. A common limitation in DPOAE i/o measurement is low signal-to-noise ratio (SNR), especially at low primary levels. We hypothesized that a peak-based method identifying a high-SNR f2 near the conventional frequency would improve the reliability and robustness of cochlear compression estimates by leveraging higher SNR at these peaks. This study compared two measurement approaches for deriving DPOAE i/o functions: (1) the conventional method using standard f2 frequencies (e.g., 1000 Hz), and (2) a peak-based method targeting individually identified maxima near the same frequency. DPOAE i/o functions were modeled using a two-segment piecewise linear fit to estimate low-level and compression slopes, and the associated change point (compression threshold). Repeatability was quantified using the intraclass correlation coefficient and Bland-Altman method. Compared to the conventional method, the peak-based approach yielded significantly higher SNR and DPOAE levels across frequencies and levels. Reliability analyses indicated that the peak-based method yielded more consistent and repeatable estimates of cochlear compression. The findings provide evidence supporting the use of fine-structure peak targeting to enhance the measurement quality of DPOAE-based cochlear compression estimates.