Dynamics Based On Human Brain Networks And Its Noise Effects

With the development of modern technologies,we now have huge amount of data for brain studies.But how to reveal useful information from these data,to help us understand the mechanisms and the principles of how brain functioning,is an important and also still difficult problem.Base on real world neural data,on the one hand,we have constructed the structural network of human cerebral cortex from diffusion tensor imaging data,studied the differences between network structures and the simulated neural activities by different subjects,investigated the relationship between these aspects and subjects' cognitive abilities,and discussed the effect of noise on the brain functions.On the other hand,using rodent hippocampal local filed potential,we extracted different spatiotemporal features,and studied the aspects of neural representations,decoding algorithms and potential applications.Through these studies,we are trying to improve our understanding of neural mechanisms behind memory,learning and cognition of the brain from three aspects: the network structure,simulated neural activities and neural representations.The main investigations and results of this thesis are as follows:1.We have investigated the relationship between the network structures of human cerebral cortex and the subject's cognitive abilities(Intelligence Quotient,or IQ).The results show that the structure of human brain network is the outcome of the trade-off between efficiency and cost of the network.The preferences of this trade-off are correlated with subject's IQ,and the male and female groups show significant difference on this correlation.2.We have studied the effect of network structure and noise on the neural population activities of different regions on human cerebral cortex.The results indicate that the network structures constraints the firing order of different regions,while the existence of noise enriches the diversities of neural activities pattern,indicating that the network structure and noise are both critical to the brain function.3.The spatiotemporal patterns of rodent hippocampal local fields potentials are studied and their applications in spatial information representation are discussed.We developed many feature extraction and neural decoding methods,and tested their application in position decoding,sharp wave ripple replay trajectory reconstruction and decision prediction.An independent real-time online decoding strategy has also developed and tested.4.The study of the effect of noise and random perturbation on the synchronization of network coupled oscillators shows that weak noise could enhance the synchronization,while the strong noise could destroy it.With the increase of strength of noise or perturbation,the phase transition of the system changes from the first-order to the second-order.These results are similar to the effect of noise on neural activities on the human brain network,indicating that the effect of noise on the collective activities of the system may have a certain universality.On one hand,these results indicate that the network structure of human cerebral cortex has an important influence on its neural activities and functions,and the existence of noise may be crucial for the maintenance of brain function.On the other hand,it shows that local field potential has great potential in the study of neural representations.Combined with our feature extraction and neural decoding methods,it could be helpful to the content-based online closed-loop neural manipulation experiments.All these results provide us with new perspectives and methods to study the mechanisms and the principles of how brain functions.