Automatic minimization of eye blink artifacts using fractal dimension of independent components of multichannel EEG

Eye blink artifact is an important artifact in EEG recordings that should be corrected before any other analysis in clinical or brain computer interface purposes. This artifact cannot be removed by frequency selective filters, because of its frequency overlap with EEG. Independent component analysis (ICA) is an effective method that can separate ocular source from brain sources. The main problem in ICA is to recognize components related to ocular artifact source, automatically. In recent years, some methods have been proposed to recognize these components based on some features of independent components. In this work, we use Higuchi's fractal dimension of independent components, because of the difference between fractal structure of the ocular and brain sources. The method has been tested by EEG data recorded for diagnose attention deficit/hyperactivity disorder (ADHD) in children. The results show that the proposed method is appropriate for automatic minimization of eye blink artifact.