In recent years, the number of studies using functional magnetic resonance imaging (fMRI) on human brain activity has increased rapidly, which has become a hot topic in medical and academic fields. The autocorrelation and correlation problems in the time series of human brain activity have also become an important research direction. It is found that there are relative residuals in the time series of human brain activity processed by smoothing splines. To solve this problem, B-spline is used to smooth the curve. By choosing the right knots, a better smoothness method to process the human brain activity data is provided. In addition, the study also found that the time series of human brain activity has correlations. The multiple scans of the same person were analyzed to see if these correlations were consistent. In order to evaluate this point, correlation is used as a response variable Y and person as a factor X to fit a random effect model. By calculating the percentage of variation in Y to determine whether the scans are similar to each other. The results show that the mean-centering time series data with 0th order difference has the most consistent correlation.
Applied Statistics (MS)
Department, Program, or Center
School of Mathematical Sciences (COS)
Zhou, Xiaowen, "Autocorrelation Reduction and Consistency Analysis of Correlation for Human Brain Activity Data" (2019). Thesis. Rochester Institute of Technology. Accessed from
RIT – Main Campus