The Effects Of Long-term Abacus Training On Brain Functional Networks Based On Graph-theoretical Network Analysis

Human brain, the most complex system, is the neural center for human cognition. In the past decades, multiple models and techniques were employed to understand, develop and protect human brain. Recently,studies found that the brain organized as a network structure, with functional integration and segregation. Furthermore, several studies revealed that the architecture of brain network has important implications associated with cognitive functions, and would be reorganized by brain maturation, disease and training. Abacus-based mental calculation (AMC) is an arithmetic skill by manipulating an imaginary abacus in mind for calculation precisely and quickly. Previous studies have reported AMC training might not only affect individuals'cognition (i.e. arithmetic ability, visuospatial working memory,executive function), but also have impact on brain function and structure.However, most previous studies were based on voxel-based analysis, and hardly to explore the relationship between different brain regions.In our study, we employed graph-theoretical analysis to examine the connectivity changes induced by AMC training. And we hope our study could provide further evidence for brain plasticity induced by AMC training.Firstly:(1) In our study, participants were all children. However, children' s MRI data was sensitive to the template employed in normalization procedure. Improper template would introduce bias and error in normalization procedure. Thus, we created study-specific template based on the participants in the present study, and tested whether this study-specific template could decrease bias induced by normalization. We found although the study-specific template was based on the small sample, it could significantly reduce individual variation, and improve the robustness of results. We also proposed and tested another method to reduce individual variation that included the difference between a template to a native image as a covariate in group-wise statistics. Thus, in other parts of our study, we all created study-specific template to promote data processing.In the current work, we designed three studies to investigate the effect induced by AMC training on brain functional network. we found:(2) Topological properties in right cingulate gyrus, inferior parietal lobe and orbitofrontal cortex (medial) would be altered by long-term AMC training. One possible explanation is the function of these regions was associated closely with visuospatial function, for that during training, AMC children would prefer to employ visuospatial strategy to solve numerical problems, while general population prefer to use linguistic strategy.(3) Human brain is organized as a network with a number of subnetworks,which represent different functions. In this part of study, we found the increased intra-connections and modularity in AMC group,which reflected increased functional segregation induced by training. Moreover, AMC training had different influences on different subnetworks, such as increased mean local efficiency in visual network, and decreased mean participation coefficiency in somato-motor network. We suspected these alterations might be related with the visuospatial function employed in AMC training,and represented the enhancement of information processing efficiency in these subnetworks. Our results might provide further evidence that AMC training might affect the functional integration between different brain regions.(4) In the present study, we also suggested the transfer effect between AMC training and high-level cognitive function (i.e.executive function). We found increased functional connectivity in frontoparietal circuit in an executive-function task. The frontoparietal circuit was involved in both AMC and executive function. Thus, we suspected that AMC training might affect the connectivity in frontoparietal circuit, and in turn, this training-induced influence might affect individuals' behavioral performance and neural mechanism. These results might open a new path to explore the transfer effect between training and high-level cognitive function.