__Keywords:__superconductiv magnetometry SQUID-gradiometer digital signal processing adaptive digital filterin biomagnetic signal magnetoencephalogram power spectral density sample estimate

I.A.Sinelnikova, Ye.P.Lobkayeva

The paper shows the possibility for applying some known methods of digital signal processing to study superweak noisy nonstationary signals of a biomagnetic field, which were produced by superconducting magnetometry using a single-channel SQUID gradiometer [1] in a weakly shielded room.
The signal being measured while researching weak and superweak biomagnetic fields (10-10…10-13 Tl) by this method always contain high-amplitude noise components of industry produced magnetic fields, geomagnetic field (GMF) and noise of a measuring complex which are million times higher of a level of signal being measured. This is a considerable restriction for SQUID-magnetometry to be adopted in a clinical practice. It is considered that usage of electronic schemes of filtration of biomagnetic signals is not always possible without loss of information which is the often reason of considerable distortions or complete losses of the final results of measurements.
The mathematical processing technique has been developed. It is aimed at selecting a magnetoencephalogram (MEG) from the noisy biomagnetic field signal and estimating its dynamic parameters.A mathematical analog of a digital recursive adaptive filter has been developed. Its frequency response allows both suppressing high-amplitude nonstationary noise components caused by geomagnetic variations and a power-line frequency electromagnetic field, and eliminating the energy spreading influence on the MEG component amplitudes. It has been concluded that it is possible to reduce the time of signal accumulation and to select it by the developed technique due to decrease of the influence produced by nonstationary noise on MEG components after their removal with a mathematical adaptive filter analog. In some cases this allows using not a smoothed estimate, but restricting to a sample estimate of a spectral density.
The mathematical biosystem model has been modified to calculate and estimate the changes in amplitude-frequency and phase-frequency characteristics of MEG. The software has been developed to computerize mathematical processing of the experimental data. It has been tested on real biomagnetic cerebral activity signals of rats and individuals.
An application of a method developed to recognize a human`s or an animal`s MEG from a strongly-noised biomagnetic signal allows to involve superconducting magnetometry in encephalographic clinical and exploratory research and to estimate its advantages. In addition to the method considered the superconducting magnetometry allows to receive extra independent diagnostic information about a psychophysiological state of a human or an animal researching an organism`s response on a magnetic interaction.

References: