Deactivating cochlear implant electrodes to improve speech perception: A computational approach.

Area of Science:

• Auditory Neuroscience
• Biomedical Engineering
• Speech Processing

Background:

• Cochlear implant (CI) electrode performance varies, potentially limiting speech perception.
• Previous studies on deactivating poorly encoding electrodes yielded inconsistent results.
• Lack of standardized criteria for electrode deactivation and consideration of speech information distribution across the array.

Purpose of the Study:

• To computationally investigate the impact of electrode selection on speech perception in CI users.
• To determine optimal electrode deactivation strategies for enhancing CI speech understanding.
• To identify patterns of active electrodes that maximize speech cue discrimination.

Main Methods:

• Utilized previously validated computational models to simulate speech perception.
• Generated over 500 million predictions across all electrode combinations in a 22-electrode array.
• Simulated three hypothetical CI user groups (better, moderate, poorer performers).

Main Results:

• Model-optimized electrode deactivation outperformed simple deactivation of poorest electrodes.
• Optimal patterns predicted an average 10% increase in word scores, with some exceeding 20%.
• Optimal configurations typically used 11-19 active electrodes, maintaining 80%-100% array span.

Conclusions:

• Strategic selection of active electrodes holds significant potential for improving CI speech perception.
• Computational modeling provides a powerful tool for identifying optimal electrode configurations.
• Further research into individualized electrode selection could enhance CI efficacy.