Comparison of Spatial Filtering Methods for Ear-EEG ASSR Estimation
Anna Sergeeva, Christian Bech Christensen, Preben Kidmose · 2024
Spatial filtering methods are widely used for decoding brain responses from multichannel electroencephalography (EEG) measurements. Ear-EEG is a method for recording EEG signals from electrodes placed in-the-ear. The Auditory Steady-State Response (ASSR) is an auditory evoked potential commonly used for hearing assessment. In the current study, we compare the conventional Best Pair method for ASSR estimation with three different spatial filtering methods: the Denoising Source Separation (DSS) method, a method based on Generalized Eigen Value Decomposition (GEVD), and a method based on solving a constrained optimization problem. ASSR-recordings from 20 subjects to five different ASSR stimuli presented at three different levels were used for the empirical comparison. All spatial filtering methods resulted in a significantly larger proportions of the detected ASSRs compared to the Best Pair. However, the differences between spatial filtering methods were not statistically significant. As a real-life application example, we applied the methods for decoding ASSR from 22 subjects, using a modified speech signal as stimulus. In summary, spatial filtering proves to be an efficient tool for enhancing ASSR detection in ear-EEG. Spatial filtering methods are computationally simple, making them feasible for integration into ear-EEG devices that can be used in everyday life settings.