Research On Cortical Resting State Functional Connectivity Analysis Based On Optical Imaging

Resting state functional connectivity(RSFC)researches can analyze the functional connectivity patterns of brain network without any stimulation.It was widely used to understand the brain mechanisms of emotion,memory and neurophysiological disease.As the researches of resting state brain functional connectivity of magnetic resonance imaging were mainly based on hemodynamic signals related to neural activity,the neural structure basis and electrophysiological mechanism of RSFC,as well as the influence of neurovascular coupling(NVC)on resting state brain functional connectivity still need to be further explored.With the development of labeling technology,wide field fluorescence imaging based on genetically encoded calcium indicator GCa MP can obtain high temporospatial resolution of neural activity signals in mouse cerebral cortex,which can be used to construct functional connectivity network with cell type specificity.It provided a powerful tool for further explaining the basic problems of resting brain functional connectivity.In this study,we used cell type specific GCa MP6 s labeled mice to study the resting state functional connectivity pattern segmentation algorithm of calcium fluorescence imaging,the comparison between resting state functional connectivity and axonal projection,and the effect of NVC on resting state brain functional connectivity.The main contents included:1)Aiming at the time-consuming and unlabeled problem of brain functional connectivity modules division of high-resolution neural activity signals,a density center based fast clustering(DCBFC)method was proposed to achieve the rapid and effective segmentation of RSFC pattern of mouse cerebral cortex.The performance of DCBFC was tested by simulation data and spontaneous calcium signal experimental data of mice which express excitatory neurons(Vglut2-GCa MP6s).Compared with the common clustering methods such as k-means,hierarchical clustering and spectral clustering,DCBFC had the fastest computing speed(p < 0.003).When clustering simulation data,the adjusted rand index(ARI)of DCBFC was higher than that of k-means and spectral clustering(p < 0.01).DCBFC showed a higher silhouette index(SI)than other methods in clustering of resting state functional connectivity of mouse brain cortex(p < 0.05).2)The neuronal axonal projection density(PD)labeled with recombinant adeno-associated virus(r AAV)was compared with the RSFC pattern obtained from DCBFC.The results showed that among the functional modules which were segmented by DCBFC,RSFC corresponding to motor/primary somatosensory areas(M/S)had the highest spatial correlation coefficient(0.91)with PD,while posterior parietal association areas/primary somatosensory trunk(Pt A/SSp-tr)had the lowest(0.75).PD projections between different functional modules not only contain all significant connections of RSFC(p < 0.01),but also contain insignificant connections among RSFC(such as the connections between Pt A/SSp-tr and visual areas(V),Pt A/SSp-tr and secondary motor area(M2),barrel field(BF)and hind limb(HL),M2 and HL,and BF and V).In addition,the average connection diversity(ACD)and average connection strength(ACS)of functional connections and axonal projections between functional modules were further analyzed.The results showed that Pt A/SSp-tr had higher normalized ACS(RSFC: 1.0,PD:1.0)and ACD(RSFC: 0.61,PD: 0.85)under both RSFC and PD connection models.3)The spontaneous calcium signals,hemodynamic signals and EEG signals of the mouse cerebral cortex were recorded simultaneously by multimodal optical imaging system.The RSFC patterns of the cerebral cortex of five GCa MP6 s mice express specific types of neurons(including excitatory neurons: Thy1-GCa MP6 s and Vglut2-GCa MP6s;inhibitory neurons: Vgat-GCa MP6 s,PV-GCa MP6 s and SOM-GCa MP6s)and the effect of isoflurane anesthesia concentration on RSFC were studied.The highest spatial correlation coefficient(0.79 ± 0.03,p < 0.001)was found between the RSFC of cortical calcium signal of Thy1-GCa MP6 s and the RSFC of cortical total hemoglobin(Hb T)of all cell types mice by using cross comparison.Spike triggered averaging(STA)was used to analyze the average response of local hemodynamic signals to "neural excitation events" with different neuron types under different depth of anesthesia.The results showed that in the anterolateral cortex,the ratio of Hb T to calcium signal was higher,while the areas with lower ratio were distributed in the middle posterior cortex;The ratio of Hb T change amplitude to calcium signal amplitude of Thy1,Vglut2 and Vgat positive neurons was less than that of PV and SOM positive neurons(p < 0.001);The delay between the change of Hb T and the calcium signal of Thy1,Vglut2 and Vgat neurons was larger than that of PV and SOM neurons(p < 0.05).