Font Size: a A A

The Component Analysis Of Multi-channal Epilepsy Signal Based On Sample Entropy

Posted on:2018-03-03Degree:MasterType:Thesis
Country:ChinaCandidate:X Q WanFull Text:PDF
GTID:2334330569486537Subject:Biomedical engineering
Abstract/Summary:PDF Full Text Request
Epilepsy is a common disease of nervous system.According to statistics,there are about 1% of people suffering from epilepsy in the world.Epilepsy patients suffer from physical injuries and mental attack which seriously affects the lives of patients and their families.When the epilepsy occurs,the synchronization of the signal will change which affects the separation results of the signal.Fast independent component analysis(FastICA)has been widely used in the signal separation because of its fast convergence speed,simple calculation and other advantages.Similarly,the change of synchronization makes the complexity of the signal change,and the sample entropy can measure the complexity of the signal.Therefore,this paper combines the sample entropy with the FastICA to analyze the preictal and interictal brain data.The main work as follows:1.An independent component analysis algorithm based on sample entropy is establishedIn this study,an independent component analysis algorithm based on sample entropy is proposed.The algorithm is based on the synchronization of the signal to analysis two situations of the preictal and interictal brain data' the independent components:(a)The difference of the number of the independent components in preictal and interictal data under the condition of different time interval.(b)The difference of the sample entropy'fluctuation of the independent components in preictal and interictal brain data.The study shows that the algorithm can efficiently analyze and contrast the characteristics of the preictal and interictal brain data.It provides a foundation for exploring the epilepsy.2.The average number of independent components of the preictal data is more than that of the interictal data.And the difference of the average number of independent components between them is the most obvious at the the time interval of 30sFirst divide the preictal and interictal data which has an acquisition time of one hour into small pieces at the time interval(T)of 20 s,30s and 60 s.Then the corresponding number of independent components of these pieces is obtained.Finally,the average number of independent components can be obtained.Compare the difference of average number of independent components of the preictal and interictal data.The statistical results show that the average number of independent components of the preictal data is generally more than that of the interictal data.And the difference of the average number of independent components is the largest between preictal and interictal data whenT=30s,which is 0.13~0.14 larger than that when T=20s and 0.2 larger than that when T=60s.Next,this study explores the the change of the number of independent components in the preictal data at different time.This paper analyzes the number of independent components per 10 minutes' preictal data within 30 minutes before the epileptic seizure,and it has a comparison with the number of independent components of the interictal data of the same length.The statistical results show that the average number of independent components of the preictal data is generally more than that of the interictal data under the conditions of the same time interval,and the closer to the epilepsy seizures the preictal data is,the more the average number of independent components it will be.The difference of the average number of independent components between the preictal and interictal epilepsy data is the most obvious under the conditions of the same data when T=30s,which is 0~0.7 larger than that when T=20s and 0.2~0.8 larger than that when T=60s.3.The sample entropy' fluctuation of the independent components in preictal data is stronger than that in interictal dataThis study analyzes the mean of the entropy' fluctuation of the preictal and interictal data in four cases:(a)The mean of the sample entropy' fluctuation of all the independent components;(b)The mean of the sample entropy' fluctuation of the original data;(c)The mean of the sample entropy' fluctuation of the first three independent components;(d)The mean of the approximate entropy' fluctuation of the first three independent components.These four cases all show that the entropy' fluctuation of the independent components in preictal data is stronger than that in interictal data.The fluctuation difference between preictal and interictal data in case(b)is 0.000077~0.000416 larger than that in case(a).The fluctuation difference between preictal and interictal data in case(c)is 0.000069~0.000081 larger than that in(b).The fluctuation difference between preictal and interictal data in case(c)is 9.217E-06~0.00015 larger than that in case(d).So the difference of entropy' fluctuation between the preictal and interictal epilepsy data in case(c)is the most obvious among them.And the sample entropy is more suitable for the signal analysis of the preictal and interictal data than the approximate entropy.
Keywords/Search Tags:epilepsy, time interval, independent component analysis, statistical analysis, sample entropy
PDF Full Text Request
Related items