Research On Fluid Intelligence Based On Multimodal Brain Connectome

Intelligence represents a decisive characteristic of human beings,which is related to all aspects of our life such as study,work,economy,health.Understanding the neural mechanisms of individual intelligence differences can optimize our cognitive development.At present,people are increasingly interested in neuroimaging biomarkers that predict cognition or health outcomes at the individual level.However,there is a lot of controversy about the predictors and neural mechanisms of individual intelligence,because most studies ignore the heterogeneity of individual intelligence and only looking for a single neurobiological biomarker,or use a single neuroimaging modality,which is difficult to generalize to new individuals.In addition,the structure of human brain is extremely complex,with tens of billions of neurons interconnected,functional and structural connectivity of the whole brain produce human behavior and cognition together,so multimodal brain connectome analysis can reveal the mechanism of the brain more comprehensively and completely.This thesis mainly utilizes structural magnetic resonance imaging(s MRI),functional magnetic resonance imaging(f MRI)and diffusion tensor imaging(DTI)data to construct cortical thickness connectivity(CTC),functional connectivity(FC),fractional anisotropy connectivity(FAC).Within a multimodal connectome-based predictive modeling(Multimodal-CPM)based on connectome-based predictive modeling(CPM)to improve intelligence prediction,exploring the neural biomarkers of multimodal connectivity related to intelligence,and the biological neural mechanism of gender differences in intelligence between individuals.The contents of this research are as follows:1.Multimodal brain connectivity improves the predictive performance of intelligence.Constructed CTC,FC,FAC,and proposed a multimodal connectome-based predictive modeling(Multimodal-CPM)based on previous onnectome-based predictive modeling(CPM)aiming at the problem of how to effectively integrate the multimodal brain connectome,also compared the prediction performance of CPM and Multimodal-CPM applied to single and multimodal connectivity for fluid intelligence from different directions.The study found that Multimodal-CPM is superior in predicting intelligence when applied to multimodal brain connections,the more modal’s connectivity are included,the higher the predictive performance,and which is significantly improved compared to CPM when applied to single-modal connections,while the prediction performance of CPM in multimodal connectivity is significantly weaker than that in single modality.2.Intelligence-predictive multimodal connectivity has overlapping and unique complementary characteristics.Applied multimodal-CPM to the combination of FC,FAC and CTC,calculated the consensus connectivity that be selected in each iteration,and explored the neurobiological characteristics behind the intelligence related multimodal consensus connectivity.The research has obtained overlapping information of consensus connectivity of each modality: FC,FAC,and CTC are collectively connected to the prefrontal lobe,occipital lobe,and limbic lobe.And the unique complementary information of consensus connectivity: FC connected to in the cerebellum,FAC connected to the motor cortex and subcortex,CTC connected to the insula and temporal lobe,FC,FAC connected to the parietal lobe,these are the unique intelligence-predictive edges of each lobe.3.Research on gender differences in intelligence.Applied Multimodal-CPM to male and female’s multimodal consensus connectivity seperately,the results showed that FC contributed more to female’s intelligence prediction than male’s,CTC contributed more to male’s intelligence prediction than female’s;The number and predictive contribution weight of FC,FAC and CTC connected to occipital lobe in female were lower than those in male,and the predictive contribution weight of FC,FAC and CTC connected to prefrontal lobe and motor cortex in female were higher than those in male;The number and predictive weight of FC connected to the default network,visual A in male is far greater than female,while which connected to the visual association network,cerebellar network,salience network in female is far greater than male.