Using independent component analysis to remove artifact from electroencephalographic measured during stuttered speech

Area of Science:

• Neuroscience
• Speech-Language Pathology

Background:

• Electro-encephalographic (EEG) signals offer insights into neural processing related to stuttering.
• Previous EEG studies are limited by significant artifact during natural speech, particularly in individuals who stutter.

Purpose of the Study:

• To demonstrate the efficacy of independent component analysis (ICA) in removing artifacts from EEG data.
• To enable the study of EEG activity during stuttered speech in children.

Main Methods:

• Recorded EEG data from 16 male children who stutter and 16 controls during a reading task.
• Applied ICA to artifact-laden EEG recordings.
• Compared EEG noise levels during stuttered versus non-stuttered speech.

Main Results:

• EEG recorded during stuttered speech exhibited significantly higher noise levels compared to non-stuttered speech (p < 0.01).
• ICA successfully and effectively removed artifacts from EEG data in all participants (p < 0.01).
• Case study illustrates ICA's capability to eliminate dominant artifacts hindering EEG analysis.

Conclusions:

• ICA is a viable method for artifact removal in EEG recordings of stuttering children.
• This technique facilitates the investigation of neural correlates of stuttering during natural speech production.