Comparison of Speech Understanding Between Tonotopy-based Fitting and Setting Based on Evolutionary Algorithms

Introduction: Cochlear implantation allows the rehabilitation of profound bilateral deafness, restoring speech perception and verbal communication when the traditional hearing aid no longer provides satisfactory hearing gain. A cochlear implant includes an electrode array and its functioning is based on the principle of cochlear tonotopy: each electrode encodes a frequency spectrum according to its position in the cochlea (high frequencies are assigned to the basal electrodes and low frequencies to the apical electrodes). The cochlear implant thus breaks down the frequency spectrum into a number of frequency bands via bandpass filters corresponding to the number of electrodes in the implant. During the fitting these bands can be modified by the audiologist. The fitting software developed by the manufacturers proposed a default fitting with a lower limit between 100 and 250 Hz according to the brands and an upper limit of about 8500 Hz. The frequency bands assigned to each electrode follow a logarithmic scale with the high frequencies for the basal electrodes and the low frequencies for the apical electrodes. This distribution takes into account the number of active electrodes but does not take into account the anatomy and the natural cochlear tonotopy specific to each patient. Several studies have analyzed the anatomical variations of the cochlear dimensions: size of the cochlea and the ratio between the contact surfaces of the electrodes with the cochlea are variable from one patient to another. The insertion depth during surgery is also variable due to parameters related to the patients as well as to the operator, which seems to impact the understanding of speech in noise. Mathematical algorithms have recently been developed to estimate the cochlear tonotopy of each patient from a CT scan assessment. CT imaging of the implanted ear combined with 3D reconstruction software, provides cochlear length measurements Using this approach it is possible to measure the position of each electrode relative to the cochlear apex. Recently, MED-EL (Austria) has developed a new approach based on CT-scan and tuning of the frequencies associated with each electrode using anatomical information of position of the electrodes in the cochlea: this fitting is called anatomy-based fitting.

A tonotopy-based fitting allows for better results in pitch perception or speech perception in noise or musical perception. However, there is a large variability in the results observed between patients. This variability could be due to the individual characteristics of the listeners (neuronal survival, current propagation and spread, duration of deafness, lack of cortical plasticity).

A modification of frequency allocations in the setting could take into account this individual variability and improve speech discrimination and music perception. Saadoun et al. (2022) studied on 27 subjects the effect of an adjustment of frequency allocation by an evolutionary algorithm (EA) approach and showed an improvement in speech perception in noise compared to a default setting (not based on tonotopy) for patients implanted for more than 6 months in a bimodal situation (with a contralateral prosthesis). We therefore propose to use evolutionary algorithms to modify the tonotopy-based fitting and improve the perception in noise of implanted patients.

Main objective:

Compare speech recognition in noise with tonotopic setting (FS4T) and tonotopic fitting modified by evolutionary algorithm (EAFS4T) in adult patients implanted for 6 months or more with a MED-EL cochlear implant with FS4T.

Secondary objectives:

Comparison of FS4T and EAFS4T settings

• for speech recognition in quiet
• for subjective auditory spatial perception
• for subjective auditory and musical perception

Plan of the study: It is a prospective open monocentric transversal study.