Epileptic intracranial electroencephalography signal warning method based on deep convolutional attention network

Abstract

The invention provides an epileptic intracranial electroencephalography signal warning method based on a deep convolutional attention network. According to the method, through acquiring an intracranial electroencephalography (iEEG) signal at an ANT (Anterior Nucleus of the Thalamus) of an epilepsy patient for analysis, extracting and fusing a multi-scale time sequence characteristic and a multi-spectrum characteristic of the iEEG signal, simultaneously adopting an attention mechanism, and concerning the most significant features of the iEEG signal in an epileptic seizure period, the warning accuracy of the network is obviously improved. In one embodiment of the invention, verification is carried out on an iEEG signal data set containing five epilepsy patients, the average epilepsy warning accuracy (Sensitivity, Sn) of a single patient can reach 95.0%, the false predicting rate (FPR) per hour is less than 0.15, the warning effect and the model generalization ability are superior to those of an existing epilepsy warning method, accurate and rapid warning of epilepsy is realized, and thus, the method is of great significance in diagnosis and nerve regulation on clinical epilepsy diseases.

Description

technical field [0001] The present invention provides a method for early warning of epilepsy intracranial electroencephalography (iEEG) based on a deep convolutional attention network. A new analysis approach is proposed, which belongs to the technical fields of signal processing and pattern recognition. Background technique [0002] Epilepsy is a common neurological disorder, affecting more than 40 million patients in the world and seriously endangering human health. At present, EEG signals are an important basis for the diagnosis and treatment of epilepsy, mainly relying on doctors to observe patients' EEG for visual detection. However, the diagnosis of epilepsy through visual detection is relatively subjective, and the burden on doctors is heavy, and it is difficult to regulate epilepsy in time when patients have seizures. Therefore, the individualized early warning technology of epilepsy EEG signals is very important, which can improve the efficiency of epilepsy diagno...

AI Technical Summary

Problems solved by technology

Although EEG signals have been widely used in the diagnosis of epilepsy, they are easily affected by various bioelectrical noises. How to exclude the influence of noise on EEG signals to accurately analyze and interpret brain activity is a major challenge.

Images