The invention belongs to the technical field of neural information, relates to an electroencephalogram
signal extraction method and specifically provides a method for electroencephalogram
signal extraction under a magnetic
resonance environment. The method is used for improving efficiency and precision of extracting weak electroencephalogram signals, wide-band electroencephalogram information is reserved so that the electroencephalogram information can be widely used for clinical application and scientific research. According to the method, first electroencephalogram signals free of magnetic
resonance interference are pre-recorded, the electroencephalogram signals under the magnetic
resonance environment are recorded, magnetic resonance gradient noises and
power line interference are sequentially removed through two-stage self-adaption
noise offset, finally high-frequency noises are removed through a
smoothing filter, and high-quality electroencephalogram signals are obtained. According to the electroencephalogram
signal extraction method, the magnetic resonance gradient noises, the
power line interference and the high-frequency noises of an instrument can be effectively removed, the high-quality electroencephalogram signals are extracted, and the
signal extraction efficiency and precision are greatly improved.