In the statistical analysis of functional brain imaging data, regression analysis and cross correlation analysis between time series data on each grid point have been widely used. The results can be graphically represented as an activation map on an anatomical image, but only activation signal, whose temporal pattern resembles the pre-defined reference function, can be detected. In the present study, we propose a fusion method comprising innovation approach in time series analysis and statistical test. Autoregressive (AR) models were fitted to time series data of each pixel for the range sufficiently before the state transition. Then, the remaining time series data were filtered using these AR parameters to obtain its innovation (filter output). The proposed method could extract brain neural activation as a phase transition of dynamics in the system without employing external information such as a reference function. The activation could be detected as temporal transitions of statistical test values. We evaluated this method by applying to optical imaging data of the respiratory related neuronal activities in the rat brainstem and demonstrated that our method can detect respiratory related spatio-temporal activation profiles in the brain.
Keywords: Innovation approach; Time series analysis; Biomedical optical imaging
Biography: He received his Ph.D. of statistical science from the Graduate University for Advanced Studies, Japan, in 2001. He worked as a research fellow in Tokyo Institute of Psychiatry (2001) and RIKEN Brain Science Institute (2001 - 2005). From 2005 to 2009 he was an assistant professor of the Department of Medical Engineering at the Chiba University, Japan. Currently He is an associate professor of the Department of Statistical Modeling at the Institute of Statistical Mathematics, Japan. His research interests are in brain signal analysis, spatio-temporal modeling and neuronal modeling.