The Classification Research Of Mental Disorders Based On Complexity Science And FMRI

Objective:Mental disorder is a kind of disease which widely affects patients' emotion,cognitive function and social interaction.It brings a heavy burden to patients and their families,and it is of great significance to accurate diagnosis and early intervention for mental diseases.However,at present,the diagnosis of all kinds of mental diseases has obvious symptomatic orientation,lack of quantitative physiological index and low diagnostic recognition rate,which leads to the delay of the disease.The purpose of this study is to analyze the brain activity of the patients with depression,schizophrenia,bipolar disorder and hyperactivity at resting state and task state using functional magnetic resonance imaging(fMRI),fractal and chaos theory,so as to explore the pathogenesis and provide effective biomarker for accurate diagnosis of mental illness.Materials and methods:The experimental design mainly focus on four types of mental disorders,including major depression disorder,schizophrenia,bipolar disorder and hyperactivity disorder.24 depressive patients and 15 healthy controls were recruited in the stress response experiment carried out at the University of Cincinnati medical research center.44 patients with schizophrenia,44 patients with bipolar disorder,34 patients with ADHD and 77 health controls were recruited in the experiment carried out at the University of California at Los Angeles.The experiment including resting state and paired word encoding,paired word retrieval,spacial working memory,task switching,balloon analog risk taking,breath holding and stop signal task.In each experiment,fMRI data were collected,and preprocessed using AFNI and fmriprep software,and credible experimental data were obtained.The AFA method was used to remove the global trend of the voxel time sequence,Hurst parameter was calculated by investigating the power law relationship between the residual and the filtering scales and serve as characterizing the long range continuity of brain activity.The SDLE method was used to investigate the divergence of the initial interference errors with the time evolution and characterize the initial value sensitivity of brain activity.Taking AFA and SDLE analysis results as characteristics,support vector machine was used to classify and predict fMRI data of different object groups.Results:1.AFA findings:Hurst parameter of resting fMRI data in schizophrenia was significantly lower than healthy control both at whole brain level and regional level,indicating decreased long range persistence of spontaneous brain activity of schizophrenic patients.The Hurst parameter of resting state in ADHD was significantly lower at left precentral gyrus,left middle frontal gyrus,right olfactory cortex etc.The Hurst parameter of depressed patients in the whole brain was reduced both in stress task and control trial,indicating the long range persistence of brain activity of depressive patients were reduced.In the cognitive task state,all kinds of mental diseases presented altered long range persistence in the corresponding brain tissue,such as in the task of paired words encoding,the upper and the rectus muscles in schizophrenia appear to be reduced persistence,while Bipolar patients showed increased Hurst at whole brain.During the breath holding task,there was no significant difference in the long range persistence of brain activity between all mental disorder groups and healthy control group.2.SDLE findings:all kinds of mental disorders have different chaotic characteristics in the corresponding brain regions,such as,at resting state,the sensitivity of the spontaneous activity of brain activity in bipolar disorder and schizophrenia patients is significantly higher than that of healthy controls.3.SVM results:taking AFA and SDLE results as characteristics,SVM-RBF was used to classify and predict fMRI data.At resting state,the classification accuracy of fMRI data of schizophrenia and healthy controls reached 83%.Note:the double sample T-TEST was used for statistical test with P value less than 0.05 indicating significant difference.Conclusion:FRMI has the complex characteristics of fractal and chaos.Using complex scientific methods such as AFA and SDLE,the nonlinear characteristics and differences of brain activity of various mental diseases can be effectively depicted,and the biological mechanism of the disease is revealed from the dynamic point of view,and the basis for clinical diagnosis is provided.