Study On Brain Functional Network Of Resting EEG In Heroin Addicts

At present,heroin addiction has a serious impact on China's public health and poses a great threat to the stability of families and society.The long-term abuse of heroin and other drugs has caused irreversible damage to the brain function of addicts,making them have a significant difference in cognition and behavior from healthy people.As we all know,the brain is responsible for cognition.Each brain region has its own specific function,and different brain regions are connected to each other to coordinate and form an extremely complex network.In this paper,we analyzed the difference in brain activity between addicted and healthy subjects from the perspective of brain function network,aiming to provide objective diagnosis basis for heroin addiction.In this paper,the magnitude-squared coherence and phase lag index were used as the connection indicators to construct the unweighted brain network and the weighted brain network for addicted and healthy subjects.Then,the two brain functional networks were used to deeply analyze the differences in connectivity and topological properties of the two groups of subjects and find the abnormal brain regions of the addicted subjects.Finally,we used the improved P-reliefF algorithm to select brain network attribute features,and used the SVM classifier to classify the two groups of subjects.The main conclusions of this paper are as follows:(1)We collected resting EEG signals from 22 addicted subjects and 22 healthy subjects,and calculated the correlation between 63 electrodes.By selecting thresholds through network sparsity,we constructed unweighted brain functional networks of two groups of subjects,then analyzed network attributes.The analysis results indicated that in the ?1 frequency band,the global coherence and brain network attribute values showed significant differences between the groups.Compared with the healthy group,the average clustering coefficient,average node degree,and small-world network attribute values of the addiction group decreased,and the characteristic path length and average betweenness increased.The difference in the topological characteristics of these networks indicated that the integrity of brain function in the addiction group was destroyed.Through the analysis of connectivity and local attributes,it was found that the addiction group and the healthy group showed differences in the frontal lobe,parietal lobe,and occipital lobe,of which the frontal lobe was the most significant.(2)We use 56 ROIs as network nodes,and use the phase lag index as the connection index.By selecting a certain threshold,we constructed a weighted brain functional network of the addiction group and the healthy group and analyzed the network attributes.In the ?1 band,the results of the global phase lag index and network attribute analysis are basically consistent with the unweighted network.The analysis results of connectivity and local attributes indicated that the two groups of subjects showed differences in the frontal lobe,parietal lobe,occipital lobe and temporal lobe,of which the frontal lobe was the main difference brain area.In the two brain networks constructed in this paper,the addiction group showed obvious abnormalities in the frontal lobe identically.(3)In this paper,through the improved reliefF algorithm,the k value is dynamically selected and the Pearson correlation coefficient is used to remove redundant features to achieve feature selection,and then the SVM classifier is used to classify the addiction group and the healthy group.Compared with the original reliefF algorithm,the improved P-reliefF algorithm improved the classification accuracy by an average of 5.25%,and the highest classification accuracy can reach 92.75%.It was found that the classification accuracy of clustering coefficients and mixed features in the weighted brain network is higher than that of the unweighted brain network.From the classification results,the highest classification accuracy rate of both brain networks is more than 87%,which also objectively proves that brain function network is a reliable method to analyze brain functional changes in heroin addicted patients.