| Background and purposeSchizophrenia is a common severe mental disease characterized by abnormal mental activity and behavior.Auditory hallucination is one of the main symptoms of schizophrenia,since auditory hallucination seriously affects the quality of life of schizophrenia patients,and as many as 25% of patients with auditory hallucination using traditional anti-schizophrenia treatment is ineffective.Therefore,it is very important to understand the neurophysiological mechanism of auditory hallucinations in schizophrenia.In recent years,to further explore the changes of brain functional connectivity,researchers have begun to estimate the time-varying functional connectivity,called dynamic functional network connectivity(dFNC),on a time scale.So far,there are few studies focus on dynamic functional connections of the first episode schizophrenia patients with and without auditory hallucination,and whether there are changes in the dynamic functional connection metrics in the onset of schizophrenia of patients with and without auditory hallucination has not been determined.Besides,it is not clear whether the cognition changes are the same between patients with and without auditory hallucination,and the relationship between dynamic functional connectivity metrics and cognition is unclear.In this research,we use the independent component analysis(ICA)approach to study the difference of dFNC metrics between first-episode schizophrenia patients with and without hallucination and the correlation between dFNC metrics and cognition,for the purpose of searching the neuroimaging markers of hallucination and the potential neuroimaging mechanism for the change of cognition.Materials and methods1.All subjects were scaned using GE Discovery 750 3.0T MR scanner,the scaning sequence including conventional MRI head scan and resting BOLD sequence.The subjects of this study were 148 patients with first-episode untreated schizophrenia(78 patients with auditory hallucinations and 70 patients without)and 129 healthy controls matched by sex,age,and years of education.We collected clinical information of all subjects,including basic information,PANSS scale evaluation of patients,and cognitive function evaluation of patients and healthy controls.2.DPARSFA is used to preprocess the resting state fMRI data,steps including excluding time points,time correction,head movement correction,spacial normalization and spacial smoothing.GIFT software was used to conduct principal component analysis and group independent component analysis on the preprocessed fMRI data,and 35 independent components(ICNs)were selected from 80 independent components as the intrinsic connection networks.The sliding time window technology was used to analyze the dFNC,and the dFNC metrics(mean dwell time,fraction of time,numbers of transition)was obtained.3.SPSS 21.0 software was used to compare the dFNC metrics between the first schizophrenia and the healthy control,as well as the auditory hallucination group,non-auditory hallucination group and the normal control group.SPSS and Matlab platforms were used to analyze the correlation between dFNC metrics and PANSS scale score and cognitive function measurement score.4.Finally,different sliding window sizes and clustering numbers were used for dFNC analysis,and the above statistical analysis steps were repeated,then the results under different conditions were compared to verify the reliability of the results.All statistical analysis were performed with the significance level of α=0.05.Results1.Clinical data analysis showed that there was no statistical difference in PANSS scale score between the auditory hallucination group and the non-auditory hallucination group(P > 0.05).Compared to healthy control group,the auditory hallucination group and non-auditory hallucination group showed impaired performance on six cognitive domains(information processing speed,attention and alertness,working memory,word learning,visual memory,reasoning and problem solving)(P < 0.05),but patients with and without hallucination showed no statistical difference across all cognitive domains(P > 0.05).2.The results of dFNC analysis showed that:(1)in state1,mean dwell time and fraction time of all schizophrenia patients were higher than healthy controls(P < 0.05),(2)in state1,mean dwell time and fraction time of non-auditory hallucination group were higher than healthy control group(P < 0.05),(3)in state2,fraction time of non-auditory hallucination group were lower than healthy control group(P < 0.05),(4)in state3,mean dwell time and fraction time of non-auditory hallucination group were lower than healthy control group(P < 0.05).There was no statistical difference of dFNC metrics between patients with and without hallucination(P>0.05).3.Comparison of functional connectivity(FC)modes of different states showed that: in state1 and state3,the FC within and between networks of first-episode schizophrenia patients was weaker compared with the healthy controls;in state3,the FC between default mode network(DMN)and precuneus of the auditory hallucination group was stronger compared with the non-auditory hallucination group.4.The correlation analysis between the clinical index and the dFNC metrics showed that the general pathological score(G)was positively correlated with the fraction time(r=0.215,P=0.021)and the mean dwell time of state1(r=0.214,P=0.021).PANSS total score(T)was positively correlated with fraction time(r=0.234,P=0.011)and mean dwell time of state1(r=0.240,P=0.009).Working memory was positively correlated with fraction time(r=0.211,P=0.017)and mean dwell time of state3(r=0.217,P=0.014).Word learning was positively correlated with fraction time of state3(r=0.177,P=0.046).Visual memory was negatively correlated with fraction time of state1(r=-0.192,P=0.030).conclusions1.The FC of the brain are constantly changing,and the first-episode schizophrenia patients spend long duration in the weak FC state.2.The enhanced FC between DMN and precuneus may result in auditory hallucination.3.The damage of the whole brain dFNC of non-auditory hallucination patients were more significant than that of patients with hallucination.4.The longer the weak FC state,the weaker the visual memory ability,and the more serious the pathological symptoms of first-episode schizophrenia patients. |