Font Size: a A A

Application Of Non-linear Dynamics Analysis Of Eeg For The Study Of Cognitive Function

Posted on:2003-05-24Degree:DoctorType:Dissertation
Country:ChinaCandidate:D Y WuFull Text:PDF
GTID:1104360092955186Subject:Neurology
Abstract/Summary:PDF Full Text Request
OBJECTIVES: In order to understand the changes of non-linear dynamics properties of EEG under different mental tasks and investigate the application of non-linear dynamic analysis for cognitive function study.METHODS: EEG was recorded in 30 healthy university students under seven states: (1)eyes closed, (2)eyes open, (3)mental arithmetic with eyes closed, (4)digit span test with eyes closed, (5)vocabulary association test with eyes closed, (6)graphics reasoning test with eyes open, (7)orientation test with eyes closed. We calculated correlation dimension (D2) and point-wise correlation dimension (PD2) for all subjects. For 5 randomized selected subjects, we performed a surrogate data test and calculated Lyapunov exponent (L1) and Kolmogorov entropy (K2). We also performed a dynamic analysis of D2 or PD2 topographics maps for different states.RESULTS: 1. For all states we tested, we could find a non-integer, fractal dimension, a positive L1 and a positive K2.2. There were significant differences of D2 and PD2 between original EEG data and phase randomization surrogate data.3. The spearman correlation coefficient between D2 and PD2 was 0.526.The result of linear regression analysis showed that there was correlation between D2 and PD2. The variation coefficient of PD2 was greater than that of D2.4. Different regions, different states, individuals and different time all could affect the results of D2 and PD2.5. D2 and PD2 increased during eyes open compared to eyes closed. D2 and PD2 increased during mental arithmetic task, digit span task, vocabulary association task and orientation task, compared to eyes closed state. D2 and PD2 increased significantly during graphics reasoning task compared to eyes open state. T3 and T5 kept higher for all states compared to average level. Since eyes movement frequently affected FP1 and FP2, the results of them were trustless. D2 and PD2 increased at all regions during five mental tasks compared to rest states. In graphics reasoning task, we found D2 and PD2 of left hemisphere were greater than those of right one and a significant increase at F3, C3, P3 and O1.6. At rest states, D2 and PD2 topographics maps kept relatively invariable, T3, T4 and T5 kept relatively higher, while O1 and O2 kept relatively lower(compared to average level). There was an overall increase of D2 and PD2 under eyes open state. Through dynamic analysis of D2 and PD2 topographics maps, there might be a working circuit from both temporal lobes to C3, F7(or F3) under mental arithmetic task. D2 and PD2 increased at T3, T4, F7, F8, F3 and F4 under digit span task. Under vocabulary association task, we could find that D2 and PD2 increased at both temporal lobes, then spread rapidly to wide regions of brain. Under graphics reasoning task, there might also be a working circuit from T3,T5, O1 and T6 to F7, F3, C3 and P3. Orientation task might have a correlation with wide regions of brain.7. Under eyes closed state and graphics reasoning task, the results of compressed spectrum array (CSA) showed that in both states evidence exists for the presence of γ frequency band. Under graphics reasoning task, the power values of γ band increased significantly.CONCLUSION: 1. The result of surrogate data test rejects the null hypothesis that both the original EEG data and the surrogate data can be described by a linear model, which argues for the presence of non-linearity in the EEG for all states we tested.2. For all states we tested, there are a non-integer, fractal dimension, a positive L1 and a positive K2.3. D2 and PD2 have a significant correlation and a same trend. D2 is steadier than PD2 and PD2 is more sensitive than D2. PD2 can track the uncertainty of EEG data. From the point of view of application, the calculation of PD2 is faster than D2, thus PD2 may be more suitable for actual work.4. Different regions, different states, individuals and different time all could affect the results of D2 and PD2. The phenomenon that the values of D2 and PD2 fluctuate with...
Keywords/Search Tags:EEG, Non-linear dynamics, Correlation dimension, Point-wise correlation dimension, Cognitive function Topographics maps, Non-linear dynamic analysis
PDF Full Text Request
Related items