Article
Investigations of pre-processing algorithm in Advanced Bionics cochlear implantat systems
Search Medline for
Authors
Published: | July 6, 2010 |
---|
Outline
Text
Background: Since a long time research have taken place on implementations of signal-pre-processing algorithms for hearing aids. But until now these experience are not considered in the cochlear implant systems.
Method: Three study conditions were evaluated in a cross-over study: the signal enhancement algorithm ClearVoice as well as a multi-channel automatic gain control (AGC) with two or four channels, respectively. Subjects received each condition for one month. Speech performance tests in quiet as well as with various types of noise were performed during each study appointment. During the first visit, the signal to noise ratio (SNR) was adjusted individually for a favourable working point within the sensitive area of the test. During the further appointments the SNR was not altered to achieve comparable results. 15 cochlear implant users shall be recruited. So far 10 cochlear implant users have already started.
Results: In stationary noise the ClearVoice algorithm improved speech performance while in simulated multi-talker babble only two of the six tested subjects showed an advantage. The absolute improvement correlated with the individual SNR indicating that poorer users were more likely to benefit from the new algorithm. The data of all subjects of the multi-channel AGC showed no significant difference to the results of the clinical progam at all kinds of noise. But about half of the subjects achieved considerable improvement.
Conclusion: Typically hearing aid users are able to understand speech even at negative signal to noise ratios while most cochlear implant (CI) users need a signal which is louder than noise. Therefore, pre-processing algorithms which are common in hearing aids may have a benefit for CI users and should be reviewed. Our results demonstrate the potential for pre-processing algorithms in CI processors. However, some algorithms seem to be more beneficial and may require less individual optimisation than others.