Motor imagery double-layer classification identification method with low false triggering rate

Abstract

The invention discloses a motor imagery double-layer classification identification method with low false triggering rate. The identification method comprises following steps: acquiring 64 lead EEG signals of one subject in different task conditions; performing band-pass filtering, down-sampling and data length interception on the 64 lead EEG signals to obtain preprocessed EEG signals; processing the EEG signals through a CSP algorithm and with the cooperation of a sliding time window method to obtain EEG signal characteristics used for later modeling; and constructing a double-layer classifier by employing the EEG signal characteristics of different categories, and accomplishing identification of the same sample through twice of discrimination to improve the overall classification identification capability. According to the method, the false triggering rate of a MI-BCI system can be effectively reduced, the overall identification performance of the system is enhanced, key problems in the research of the conventional MI-BCI system are improved, and if the algorithm is applied, considerable social benefit and economic benefit can be obtained.

Description

technical field [0001] The invention relates to the field of classification recognition, in particular to a motor imagery double-layer classification recognition method with low false trigger rate. The present invention proposes a novel method for improving the overall classification and recognition ability of a Motor Imagery Brain Computer Interface (MI-BCI) system. Background technique [0002] Brain Computer Interface (BCI) technology is a relatively popular research direction in the field of neural engineering, and has a history of nearly 50 years. Among them, MI-BCI is a rising star in BCI research, and it is the most meaningful one among the technologies currently used to control external neural prosthesis. [0003] When the human body completes the actual action or imagines the action, the activity state of the sensorimotor area of ​​the cerebral cortex will change, and the EEG signals of certain specific frequencies (mainly concentrated in the α frequency band and β...

AI Technical Summary

Problems solved by technology

[0007] 1. The completion rate of the target task is low, and the correct command output cannot be realized to control the external equipment to complete the preset task;

[0008] 2. It affects the operator's judgment of the degree of participation of system users, and thus cannot effectively evaluate the performance of the system;

[0009] 3. It is easy for users to form wrong motor imagery habits, which will bring hidden dangers to the use of other similar equipment in the future;

[0010] 4. When it is applied in the field of rehabilitation, it will waste the precious rehabilitation treatment time of patients and delay the rehabilitation process

At present, most MI-BCI research focuses on the improvement of the system's classification accuracy of the target task, but no research has been found to explicitly raise the problem of false triggering (the first type of error) and overcome it.

Images