Imaging Markers Of Drug Addiction Based On Multimodal Magnetic Resonance Imaging

Drug addiction,a kind of chronic and relapsing brain disease,has become one of the main factors of seriously endangering people’s health and community stability.There is currently no effective approach for the prevention and treatment of drug addiction,especially the problem of relapse.Magnetic resonance imaging(MRI)is one of the main tools for the study of addiction,but most of the existing researches have focused on the use of unimodal MRI images to explore structural or functional changes in addicted brains,and there is a major trend to combine multimodal imaging information to find reliable imaging markers of addiction.Therefore,this paper proposes to combine structural MRI,diffusion kurtosis imaging(DKI)and resting-state functional MRI(rs-f MRI)to explore the neural mechanisms of drug addiction.The main research contents of this paper are as follows:(1)Using structural MRI and DKI imaging data,to explore the differences in gray matter(GM)regions between heroin addicts(HAs)and healthy controls(HCs),and to define imaging markers of drug addiction.In this study,we used voxel-based morphometry(VBM),DKI analysis,and two-sample t-test to explore differences between HAs and HCs brains.In addition,cognitive tests were performed on all subjects,and the associations between image parameters and cognitive scores were analyzed.The results of the experiment showed that HAs in the anterior cingulate cortex(ACC),middle cingulate gyrus(MCC),supplementary motor area(SMA),precuneus(PCE),paracentral lobule(PCT)and the medial superior frontal gyrus(MSFG)showed macroscopic and microstructural differences.Moreover,our results revealed that changes in brain structure are closely related to the cognitive deficits of patients.(2)Using DKI and rs-f MRI,to explore the differences in brain network between HAs and HCs,and to explore the influence of drug addiction from the perspective of brain connectivity.In this study,structural and functional connectivity networks were constructed based on DKI and rs-f MRI,respectively,and a two-sample t-test was performed to explore differences in brain connectivity between groups.The results showed that there were structural and functional differences within and among the default mode network(DMN),executive control network(ECN),salience network(SN),and striatum network(ST)in HAs,which demonstrated that this abnormality has been shown to be closely related to the patient’s deficits in cognitive function and ability to control execution.(3)Based on the radiomics method,we extracted the radiomics features of the parametric map and constructed a logistic regression model,which takes as input these features,predicting the classification accuracy corresponding to each brain region.Then,the brain region with higher classification accuracy was defined as the imaging markers of drug addiction.The experimental results showed that the imaging markers defined by this method holds almost same tends as the results gained by our first work mentioned in(1),which further proved the correctness and reliability of the prediction of addiction image markers.