Reference noise cancellation: Optimizations and Caveats

Abstract

Reference noise cancellation algorithms theoretically allow the removal of external noise from biomagnetic measurements without attenuation of signal components in the same frequency band. The technique requires reference sensors that are insensitive to the biomagnetic signals of interest, yet subject to the same environmental noise. conditions. By estimating the correlation between reference (e.g. distant magnetometers) and measurement (e.g. gradiometer) Channels it is possible to form a prediction of the measurement Channel based on reference Channel Information; this prediction is defined as noise and subtracted from the data. The filters considered in this paper predict the measurement channel from the superposition of linearly weighted reference Channels; the weights are determined through a least squares minimization of the filter Output power.