8th Speech in Noise Workshop, 7-8 January 2016, Groningen

The development of SpiN recognition by French cochlear implant users

Chris J James(a)
Cochlear France and CHU Toulouse, FR

Chadlia Karoui(b)
Cochlear France and CHU Toulouse, FR

Mathieu Marx
CHU Toulouse, FR

Charles-Edouard Molinier
CHU Toulouse, FR

Benoit Lepage
CHU Toulouse, FR

Olivier Deguine
CHU Toulouse, FR

(a) Presenting
(b) Attending

Understanding the time-course of development of speech recognition in noise in cochlear implant users may allow us to better discover factors limiting performance and ways to improve cochlear implant design and sound coding, and to develop and customise rehabilitation strategies. Patient factors such as the history and pathology of deafness were investigated as well as CI brand and type and position of electrode array.
We discuss longitudinal results for the entire sequence of 136 adults unilaterally implanted in our centre in Toulouse between 2010 and 2014. These cochlear implant (CI) users were followed from the day of activation to 12 months post-activation. French MBAA2 lists of 15 sentences containing 100 words were used for all tests with eight-talker babble noise. At each visit CI users were tested using implant alone with one list in quiet and at fixed SNRs (10, 5, 2.5 and 0 dB). SNRs were reduced until word-in-sentence scores were <50% correct.

A program of post-operative CT imaging was introduced to determine the position of implanted electrode arrays. Complete data-sets were available for sub-group of seventy-eight CI users.

Mean scores in quiet after one month’s use of the CI were ~70% correct, and in 10 dB SNR ~55%, with a large range (0 to 100). Sentence scores progressed ~10% points per three months from one to nine months of use. After twelve months most CI users developed high scores in quiet and >50% of users obtained scores >80% correct in 10 dB SNR. Multivariate analysis of the subgroup with CT data showed that implant brand had some effect on scores in quiet and in noise; however one of the three brand sub-groups was small (N=5) and there was strong co-variation of brand with insertion depth. There was no significant difference in scores between CI users due to perimodiolar versus straight design or the presence of scala dislocation. Further step-wise multiple regression revealed cause of deafness to be a significant factor: Subjects with any particular pathology could obtain high scores in quiet whereas those with a history of chronic otitis, Meniere’s, and otosclerosis generally had lower scores in noise (p<0.05). Those with late onset genetically-related deafness had wide-ranging scores in noise compared to those with deafness caused by meningitis, noise exposure, ototoxicity or trauma (all >75% correct in 5 dB SNR). Some pathologies were limited to certain age groups which may have contributed to some of these differences.

Last modified 2016-05-12 14:22:09