Research On Prefrontal Brain Network Of FNIRS Signal For Depressive Disorder Patients

Depressive disorder is the most common mental disorder in today's society,with high incidence and high recurrence.Therefore,how to identify depressive disorder quickly and effectively is the key problem that needs to be solved urgently.In recent decades,with the worldwide attention to mental health,the researches on mental disorders have shown a trend of multi-angle and multi-means.Many physiological signal detection methods were applied to the screening and diagnosis of mental disorders.Among these methods,fNIRS has attracted more and more attention in the field of brain science because of its high anti-interference and safety.Traditional diagnosis of depressive disorder relies on the experience of professional doctors,mainly through screening scales,direct observation,examination and analysis to determine the condition.With the penetration of artificial intelligence in various industries,researchers try to use data as the driving force to find some relatively objective diagnosis methods for depressive disorder.In this paper,the activation of prefrontal lobe in depression patients was studied by fNIRS technique based on two experimental designs.The specific research contents are as follows:(1)Based on the characteristics of fNIRS,the resting state experimental paradigm and the task-based experimental paradigm based on emotion word Stroop effect were designed.The changes of blood oxygen concentration in the prefrontal cortex(PFC)of depression patients and normal control group were measured.Both the resting data usually used in traditional functional brain network researches and the task state data for the purpose of emotional induction and activation of brain network were obtained.(2)The variation coefficients of the channels and the experiment trials were calculated to screen the high-quality experimental data.Then we tried a variety of artifacts processing methods including principal component analysis(PCA),spline interpolation(Spline),wavelet filtering(Wavelet).After comparison,PCA method was finally adopted to de-noising.(3)According to the characteristics of emotion word Stroop effect experiment,we analyzed the behavioral differences between the depression group and the control group under different emotional stimuli.Statistical methods were used to analyze the tasks reaction time and accuracy of the two groups,the results showed that the depression patients need longer response time and lower accuracy when completing the task experiment.(4)The correlation coefficients between channels were calculated for resting state data and task state data respectively.The connection matrices were constructed by pearson correlation analysis(CORR),amplitude square coherence coefficient(COH)and phase locking value(PLV).These connection matrices contain brain connection information.The results of network analysis indicated that there are small world models in brain network of both depression patients and healthy people.In addition,network parameters such as the characteristic path length and the betweenness centrality were extracted from the connection matrices.The comparison results showed that there are certain differences between depression group and control group.(5)Combining with the network analysis results of the two groups of experiments,we extracted some features and briefly discussed the classification effect of the two groups of subjects by machine learning methods.It was found that the classification effect was better on the data of deoxygenated hemoglobin concentration(Hb R).In conclusion,this paper,starting from the experiment,researched on the PFC brain network of depression patients in resting state and task state according to the characteristics of depressive disorder.With the brain network features compared and analyzed,this study provides a new idea for the detection and diagnosis of depressive disorder to a certain extent.