MRI Based Functional Connectivity Density Approaches And Pattern Classification Prediction Of Young Smokers

Smoking addiction is one of the most serious problems in the society,which seriously harms the physical and mental health of human,especially for young people.At present,researchers use brain imaging technology to find that smoking addiction is related to the brain structural and functional changes in smokers,which is conducive to further understanding of the brain mechanism of smoking addiction.However,the neural mechanism of smoking is still unclear.Studies have shown that smoking withdrawal can induce cravings.However,few studies used functional connectivity density mapping(FCDM)method to examine the hub of brain gray matter in the abstinence induces craving of young smokers.Therefore,we employed FCDM method to compare the resting-state properties between 12-h abstinence and satiety conditions in young adult smokers.The differences of neuroimaging are not clear whether the results of differences/correlation can be applied to new samples.We use the pattern classification method in machine learning to accurately predict individual attributes.This is the second research point.The former study provides theoretical guidance for the latter research,while the latter research further promotes the former research,and the two studies are closely linked.The research contents and innovations of this paper are as follows:(1)Functional connectivity method was used to explore the brain functional change patterns of abstinence-induced craving in adolescent smokers.In this paper,gaussian kernel function is used to improve traditional linear time series,and a time series model based on gaussian kernel is proposed.Compared with the traditional model,the improved model has better fitting effect.Combined with the improved model and functional connectivity analysis method,the functional connectivity of adolescent smokers in the two states was extracted,and paired T test is conducted to obtain the difference in functional connectivity of abstinence-induced craving brain.Multiple regression analysis method is used to explore the relationship between craving and functional connectivity of adolescent smokers.This study promotes the understanding of the pattern of changes in the whole brain function during smokers’ short-term abstinence,and provides certain imaging basis for the neural mechanism of smokers’ abstinence-induced craving.(2)Pattern classification method is used to identify adolescent smokers.This paper proposes a pattern classification method based on multi-modal image data feature fusion and multi-attribute selection,and applies this method to the study of adolescent smokers.In this work,multimodal magnetic resonance imaging data of adolescent smokers are used to extract various image indexes and conduct feature fusion.Then,the low-rank self-expressed attribute selection algorithm is used to select the features.Finally,linear support vector machine is used for classification.Meanwhile,logistic regression is used to verify the robustness of linear support vector machine,whether the contributed brain areas/contributed features of the two classifiers are the same,and whether the weights of contributed features are related.The results show that the effect of fusion features is better than that of univariate analysis.The brain region that contributed the most to classification involved the brain region of the striatum-frontal circuit,which is also found in work 1,confirming the results of work 1.In conclusion,this paper used functional connectivity method to reveal that the intensity of abstinence-induced craving in smokers is related to the brain neural mechanism of cognitive control injury.In addition,the pattern classification method based on multiple attribute selection is used to classify smoking individuals.Both studies found that the brain region of the striatum-frontal circuit is an important part of the brain in smokers.