Using FMRI Brain Network Method To Study The Impact Of Training On Four Visual Cognitive Tasks

Long-term training can effectively improve people's cognitive control ability and behavior performance.This process is often accompanied by changes in brain structure and function.However,the relationship between this change and the benefits of training remains to be discovered.In recent years,with the continuous development of magnetic resonance imaging and brain network technology,more and more researchers are committed to using brain network methods to explore the impact of training on brain structure and function changes.This article will use the functional brain network method to conduct global attribute analysis and classification research on the brain networks of the subjects after two different trainings.The main work is as follows:(1)The functional magnetic resonance data(fMRI)of the control group,the memory palace training group and the N-back training group before and after training were pre-trained at rest and under four different visual cognitive tasks.We first extract the time series of each brain region according to the template,and build a functional brain network by calculating the correlation between the brain region time series.Finally,the graph theory method was used to analyze the differences of the global attributes of the functional network between the three groups of subjects before and after the training under five conditions.(2)Feature selection for each attribute of the constructed functional brain network,we use feature selection algorithm(SS_LR),random subset feature selection algorithm(RSFS),and mutual information score feature selection algorithm based on stationary selection(MI)algorithm completes feature selection,and uses nested cross-validation SVM classifier to achieve classification.Discuss the brain network classification effects of different training groups under different tasks.The results of brain network analysis showed that the control group did not reach statistically significant differences in the whole threshold range of global attributes before and after resting state,N-back task,and Stroop task.The analysis of the global attribute difference between the resting state brain network and the four task state brain networks of the memory palace training group showed that the difference before and after the test was statistically significant under some thresholds.In the N-back training group,under partial thresholding,the global attributes of resting-state,N-back task,Stroop task,and vocabulary memory task before and after the test of the brain network reached significant differences.Overall,the global efficiency,local efficiency,and average clustering coefficient of the brain network increase with the increase of thresholding.On the contrary,the small world attributes and feature path lengths have a downward trend with the increase of thresholding.The N-back training group has higher global efficiency,local efficiency,and small-world attribute values for the brain network attributes under the N-back task and the Stroop task after training than before training.The memory palace training group found a shortening of the feature path length on the attributes of the brain network before and after resting.The brain network classification results show that the control group has the worst classification effect,the memory palace training group and the N-back training group have better classification performance,and the RSFS feature selection algorithm has the best classification performance.This means that the brain network before and after the test of the training group has changed significantly,and the classification results of the resting brain network before and after the test are better and stable than the classification results of the task brain network,which shows that the brain functional changes caused by training are It can be well reflected in the resting state.Long-term training is a stable change to the changes in brain function and structure.By comparing the memory network classification results of the memory palace training group and the N-back training group before and after the Stroop task and the emotion regulation task,it can be seen that after the two training methods at the same time,the N-back training performs the Stroop task for the subjects The improvement of emotional adjustment tasks is greater than that of memory palace training,which means that the improvement effect of N-back training can be obviously transferred to Stroop tasks and emotional adjustment tasks that also require cognitive control skills.On the contrary,the effect of memory palace training is difficult to transfer Go to the Stroop task and emotion regulation task.