Electroencephalographic (EEG) recordings provide objective estimates of listeners' cortical processing of sounds and of the status of their speech perception system. For profoundly deaf listeners with cochlear implants (CIs), the applications of EEG are limited because the device adds electric artifacts to the recordings. This restricts the possibilities for the neural-based metrics of speech processing by CI users, for instance to gauge cortical reorganization due to individual's hearing loss history. This paper describes the characteristics of the CI artifact as recorded with an artificial head substitute, and reports how the artifact is affected by the properties of the acoustical input signal versus the settings of the device. METHODS: We created a brain substitute using agar that simulates the brain's conductivity, placed it in a human skull, and performed EEG recordings with CIs from three different manufacturers. As stimuli, we used simple and complex non-speech stimuli, as well as naturally produced continuous speech. We examined the effect of manipulating device settings in both controlled experimental CI configurations and real clinical maps. RESULTS: An increase in the magnitude of the stimulation current through the device settings increases also the magnitude of the artifact. The artifact recorded to speech is smaller in magnitude than for non-speech stimuli due to signal-inherent amplitude modulations. CONCLUSION: The CI EEG artifact for speech appears more difficult to detect than for simple stimuli. Since the artifact differs across CI users, due to their individual clinical maps, the method presented enables insight into the individual manifestations of the artifact.
Electroencephalographic (EEG) recordings provide objective estimates of listeners' cortical processing of sounds and of the status of their speech perception system. For profoundly deaf listeners with cochlear implants (CIs), the applications of EEG are limited because the device adds electric artifacts to the recordings. This restricts the possibilities for the neural-based metrics of speech processing by CI users, for instance to gauge cortical reorganization due to individual's hearing loss history. This paper describes the characteristics of the CI artifact as recorded with an artificial head substitute, and reports how the artifact is affected by the properties of the acoustical input signal versus the settings of the device. METHODS: We created a brain substitute using agar that simulates the brain's conductivity, placed it in a human skull, and performed EEG recordings with CIs from three different manufacturers. As stimuli, we used simple and complex non-speech stimuli, as well as naturally produced continuous speech. We examined the effect of manipulating device settings in both controlled experimental CI configurations and real clinical maps. RESULTS: An increase in the magnitude of the stimulation current through the device settings increases also the magnitude of the artifact. The artifact recorded to speech is smaller in magnitude than for non-speech stimuli due to signal-inherent amplitude modulations. CONCLUSION: The CI EEG artifact for speech appears more difficult to detect than for simple stimuli. Since the artifact differs across CI users, due to their individual clinical maps, the method presented enables insight into the individual manifestations of the artifact.