Study On Dynamic Characteristic Of Spontaneous Brain Activity And Correlation Brain Network In Patients With Parkinson’s Disease Based On Resting State Functional Magnetic Resonance Imaging

Objective This study examines abnormal dynamic characteristics of regional homogeneity(ReHo)and amplitude of low-frequency fluctuation(ALFF)and the dynamic changes of functional connectivity(FC)between altered brain regions and whole brain gray matter in Parkinson’s disease(PD)patients by resting-state fMRI.The purpose is to explore the abnormal interactions between brain networks,thereby understanding the potential mechanism of movement disorder and other related complications more deeply.Methods The resting-state fMRI data of 54 PD patients who have enrolled in the Subei People’s Hospital from August 2020 to August 2021 and 50 age-and sex-matched healthy controls(HCs)were prospectively collected.Temporal Dynamic Analysis(TDA)toolbox based on MATLAB were used for data processing.The difference of dynamic regional homogeneity(dReHo)between PD and HC groups were detected by two sample t-test.The altered brain regions were selected as seed points to calculate the functional connectivity(FC)between them and whole brain gray matter.The differences of functional connectivity variable coefficient(FCVC)between the PD group and the HC group were compared by two-sample t-test.The cognitive function of the patients was evaluated with the Mini-Mental State Examination(MMSE)scores and Montreal Cognitive Assessment(MoCA)scores.The third part of Unified Parkinson’s disease Rating Scale(UPDRS)was used to assess the patients’ motor function.Then,the correlation between the signal values of altered brain regions and clinical scale scores was analyzed.A sliding time window method was used to calculate the dynamic amplitude of low-frequency fluctuation(dALFF)while dALFF states were identified by a k-means clustering method.Then,the intergroup differences of dALFF variability and state metrics between PD and HCs were compared.The correlations between dALFF variability,states metrics and clinical scores were further analyzed.Finally,machine learning models were constructed based on ALFF feature vectors to evaluate the differentiating efficiency of each classification model to distinguish PD patients from HCs.Results Compared with HC group,PD group showed decreased dReHo in the left caudate,right putamen and bilateral lingual(P<0.05);The dReHo variability in the left caudate(r=0.374,P=0.04)and right putamen(r=0.379,P=0.03)was positively correlated with the MMSE scores.The dReHo variability in the left caudate(r=-0.446,P=0.01),right putamen(r=-0.369,P=0.04)and left lingual(r=-0.419,P=0.02)was negatively correlated with the UPDRS-Ⅲ scores.Compared with HC group,PD group showed a significant decrease in FCVC was found in the left lingual and right lingual,bilateral postcentral gyrus,right precentral gyrus and right precuneus lobe.The FCVC variance between left postcentral and left lingual was negatively correlated with the UPDRS-Ⅲ scores(r=-0.438,P=0.01).The FCVC variance between right postcentral and left lingual was positively correlated with the MoCA scores(r=0.415,P=0.02).Compared with HCs,PD patients exhibited decreased dALFF variability in the bilateral lingual gyrus,left superior frontal medial gyrus,right superior frontal gyrus and right cerebellum.The dALFF variability in the right superior frontal gyrus was significantly correlated with the MMSE scores.Three repetitive states(state Ⅰ,state Ⅱ and state Ⅲ)were identified by k-means clustering.Relative to HC group,the PD group demonstrated a shorter mean dwell time(MDT)in state Ⅱ,a longer MDT in state Ⅰ and state Ⅲ,and fewer transitions.Furthermore,the machine learning demonstrated a higher diagnostic performance of dynamic feature vector for Parkinson’s disease(AUC 0.822).Conclusion PD patients have functional abnormalities in multiple brain regions at resting state and significant time variability,which affects their movement and cognitive function.The dynamic characteristic in related brain regions could classify PD patients and healthy controls at the individual level better.