Research On Emotion Classification And Application Based On Brain Functional Connectivity Network

Long-term emotional despair can lead to mental collapse and depression,affect people’s social functions and even endanger life safety.With the development of brain computer equipment,the power improvement on computer computing and the in-depth research on emotion classification algorithms,researchers’ cognition of the emotional neural mechanism increases and the combination of emotional neural mechanism with medical health is more widely applied.Starting from the brain functional connectivity network,from single modal to multimodal,from emotion classification to methods,multimodal feature fusion methods and major depressive disorder(MDD)recognition methods,analyzes the cognitive mechanism of emotional stimulation and MDD,explore the brain functional connectivity network changes,multimodal complementary representation features and potential biomarkers of MDD recognition after emotional stimulation.The main research contents of this thesis are as follows:(1)Traditional methods are difficult to analyze the connection characteristics of brain functional connectivity network after calm and emotional stimulation,an emotion classification method based on brain functional connectivity network is proposed.This method classifies emotions based on the global and local features of brain functional connectivity network,extracts baseline and experimental signals to compare and explore whether there are different in brain functional connectivity network of clam or emotional stimulation in brain functional connectivity network.The experimental results demonstrate that after emotional stimulation,the intensity and number of brain functional connections increase.When fixed valence dimension is high or low,clustering coefficient,average shortest path length,global efficiency,local efficiency and node degree are positively correlated with the arousal degree in the arousal dimension.At meantime,the right brain connection defect is more evident with the increase of arousal degree.(2)The existing multimodal methods are difficult to express the emotion classification mechanism in different states and the multimodal complementary representation features after emotional stimulation,an emotion classification method based on the fusion of brain functional connectivity network and eye gaze(ECFCEG)is proposed.From the perspective of affective computing,five global features and two local features of brain functional connectivity network and five features of eye gaze are analyzed,the influence of different features on emotional mechanism is discussed from different perspectives.The experimental results manifest that the multimodal complementary representation features can effectively improve the accuracy of emotion classification.Meanwhile,there are defects in the right temporal lobe(RT)and the right posterior lobe(RP)in the low frequency band after emotional stimulation.(3)The existing MDD recognition algorithm based on single-layer neural network method can not simulate the stability of complex connection between brain functional regions,a MDD recognition method based on the multi-layer brain functional connectivity network(MBFCN)is proposed,and the cognitive analysis is conducted.Experimental results and cognitive analysis display that major depressive disorder recognition method based on multilayer brain functional connectivity network method contributes to improving the stability of MDD,the connection between the right prefrontal lobe and temporal lobe is defective in MDD based on the phase lag index(PLI).In addition,potential biomarkers are identified by the significance analysis of MDD features and PHQ-9.