SSDA-based electroencephalogram signal ocular artifact removal method

Abstract

The invention relates to an SSDA-based electroencephalogram signal ocular artifact removal method, which comprises the steps of S1, performing normalization processing on a pure electroencephalogram signal, and inputting the pure electroencephalogram signal into an SSDA model; s2, not training SSDA according to the minimum error between the reconstructed EEG signal and the pure electroencephalogram signal, and continuously adjusting model parameters; and S3, normalizing the electroencephalogram (EEG) signal containing the EOG artifacts, inputting the normalized electroencephalogram (EEG) signal into the trained SSDA model, and performing reverse normalization processing on output data to obtain an EOG-removed EEG. According to the method, the use of the electro-oculogram artifacts as reference signals in the electro-oculogram artifact removal process can be avoided, and the effectiveness and real-time performance during electro-oculogram artifact removal are ensured.

Description

technical field [0001] The invention belongs to the field of electroencephalogram signal processing in brain-computer interface, in particular to a method for removing electrooculogram artifacts of electroencephalogram signals based on SSDA. Background technique [0002] Currently, non-invasive research methods are commonly used in brain-computer interface research, but this method has certain drawbacks. Since the electrodes are directly attached to the surface of the cerebral cortex, the subjects will blink during the signal collection process. This can introduce electro-oculogram (EOG) artifacts. The electrical signals generated by brain activity are weak, and the existence of artifacts will cover up the real EEG signals, resulting in a sharp decline in the quality of EEG signals. Studies have shown that the oculograph artifacts and EEG signals overlap to a certain extent in certain frequency ranges, and it is not feasible to remove them directly through filters. Therefo...

AI Technical Summary

Problems solved by technology

At present, blind source separation methods such as principal component analysis (PCA) and independent component analysis (ICA) are often applied to oculoelectric artifact removal, but such methods need to analyze a large number of EEG channels, and the signal separation process is very time-consuming.

In addition, blind source separation can easily cause the loss of effective components of EEG signals and increase the similarity between channels.

With the development of neural networks, the FLN-RBF algorithm and FLNN-ANFIS were proposed. Both of these algorithms use the good approximation performance of neural networks to achieve a better effect of removing oculograph artifacts, but the two algorithms Both require additional oculoelectric electrodes to collect oculoelectric signals as network reference signals, which is not conducive to future integrated applications

Images