Simplifying cochlear implant speech processor fitting

Conventional fittings of the speech processor of a cochlear implant (CI) rely to a large extent on the implant recipient's subjective responses. For each of the 22 intracochlear electrodes the recipient has to indicate the threshold level (T-level) and comfortable loudness level (C-level) while stimulated with pulse trains. Obtaining these behavioral measurements is a time-consuming task. It requires cooperation and considerable effort of the CI recipient. Especially in adults that have been deaf for many years, and in toddlers and infants, it can be laborious to obtain these behavioral measurements. Using the thresholds of electrically evoked compound action potentials (ECAPs) in the fitting procedure allows for a more automated fitting. The profile of ECAP thresholds across the electrode array proved in a principal components analysis to be largely governed by two factors: shift and tilt. Shift represents a change in level, equal at all electrodes, and tilt represents a change in the slope of the profile. The 44 subjective responses to pulse trains in the conventional fitting have in the ECAP-based fitting been replaced by only two subjective responses to live speech: the level at which speech becomes just audible and the level at which it is comfortably loud. The ECAP-based fitting procedure proved to be (1) less strenuous for the recipient, as less subjective responses are needed, (2) faster, and (3) easier for the clinician. Speech perception scores were equal for the ECAP-based and the conventional fitting procedure. ECAP-based T-levels were much lower than conventional T-levels, whereas C-levels were almost equal in the two fitting methods. Additionally, we offered CI recipients the opportunity to optimize their ECAP-based fitting themselves by adjusting shift and tilt during everyday life. This fine-tuning by the recipients themselves bypassed the need for interpretation by the clinician of the recipient's description of his auditory percept. The recipients did not adjust the fitting to make a 'comfortable' sound, but actually tried to optimize speech perception by increasing high-frequency stimulation and hence making the sound sharper. Finally, speech perception was tested using various settings of T- and C-levels, and after eliminating soft stimuli from the speech signal. The results show that audibility is crucial; the upper part of the electrical dynamic range (EDR, the range between T- and C-level) is most important in speech understanding. Using this part of the EDR yielded the highest speech perception scores at normal and low presentation levels, compared to using the middle or the lower part of the EDR. The results show that raising T-levels does increase speech perception scores at low presentation levels of speech. Eliminating low level speech elements, up to 35 dB SPL, did not affect speech perception scores for speech presented at 60 dB SPL. This finding explains why the downward expansion of the EDR due to lower T-levels in the ECAP-based fitting did not affect speech perception at presentation levels of 55 dB SPL and up. In summary, the ECAP-based fitting procedure is fast and easy for the recipient, yielding good speech results