Clustering visualization method for epilepsy data dimension reduction based on salient point self-adaptive isometric mapping manifold

Abstract

The invention relates to a clustering visualization method for epilepsy data dimension reduction based on salient point self-adaptive isometric mapping manifold, which is used for analyzing evolutionlaws of an epilepsy patient in different periods, and specifically comprises the following steps: selecting an electroencephalogram data set of the epilepsy patient; setting various initial parametersof the salient point self-adaptive isometric mapping manifold; constructing a neighborhood map of the electroencephalogram sample; selecting significant electroencephalogram sample points, and calculating the shortest path from the significant sample points to all the sample points; updating the distance matrix by a piecewise linear adaptive adjustment mechanism; calculating a low-dimensional embedded coordinate; adaptively optimizing an embedded coordinate; and carrying out two-dimensional and three-dimensional clustering visual analysis. The invention is obvious in visualization effect andgood in clustering performance. According to the invention, professional medical staff can be assisted in marking data, and the medical staff and personnel in the cross field can conveniently analyzethe intrinsic pathogenesis of the clustering rule generated by the electroencephalogram of the epilepsy patient in different periods.

Description

technical field [0001] The invention relates to the field of cross-information medicine such as EEG clustering and visualization for dimensionality reduction of epilepsy data. Background technique [0002] As a chronic neurological disease, epilepsy seriously damages the physical and mental health of patients. The detection of epileptic EEG signals in different periods can help doctors diagnose the condition. In addition, if epileptic seizures can be predicted in time, it can significantly improve the quality of life of epilepsy patients and open up new therapeutic ideas for the prevention and treatment of epilepsy diseases. Effective feature extraction methods become the key to epilepsy detection and epilepsy prediction in different periods. [0003] Although PCA (Principal Component Analysis), LDA (Linear Discriminant Analysis) and NMF (Non-negative Matrix Factorization) have better effects in feature extraction applications, they are all linear models and can only disco...

AI Technical Summary

Problems solved by technology

[0004] The present invention is used to solve the problem that the cost of labeling the above-mentioned EEG signals is high and the clustering visualization effect of the commonly used unsupervised clustering algorithm is relatively poor. The specific technical solutions are as follows:

Images