BLASST: Band Limited Atomic Sampling With Spectral Tuning With Applications to Utility Line Noise Filtering.

Area of Science:

• Biomedical Signal Processing
• Neuroscience

Background:

• Utility line noise is a common artifact in biomedical signals, particularly electroencephalography (EEG).
• Nonstationary line noise can obscure neurological signals of interest and is challenging to remove with conventional filtering techniques.

Purpose of the Study:

• To introduce and evaluate a novel method, band limited atomic sampling with spectral tuning (BLASST), for identifying and removing nonstationary utility line noise.
• To compare the performance of BLASST against existing filtering approaches.

Main Methods:

• BLASST is an iterative algorithm that fits nonstationary line noise using Gabor atoms.
• The method utilizes surrounding frequency information for self-modulation and termination based on a convergence criterion.
• Simulated and real instances of nonstationary line noise were used to test BLASST and alternative filters.

Main Results:

• BLASST demonstrated effective fitting of line noise.
• The method showed superior preservation of local signal features compared to alternative filtering techniques.
• BLASST successfully handled highly nonstationary line noise.

Conclusions:

• BLASST offers a valuable alternative to traditional bandpass, notch, or other filtering methods for biomedical signals contaminated with nonstationary line noise.
• This method is particularly significant for EEG experiments where line noise can dominate weak neurological signals.
• A MATLAB toolbox for implementing BLASST is available.