Dynamic analysis of functional Magnetic Resonance Images time series based on wavelet decomposition

In the research of brain and cognitive science, the key problem of analyzing functional Magnetic Resonance Imaging (fMRI) data is not only to detect and locate the functional active signal accurately but also to obtain the dynamic changes of activated areas. This paper represents a novel approach to decompose the time series data in activated areas based on wavelet analysis for fMRI data processing; the general tendency and the periodic active components during fMRI experiments can be extracted with analyzing the wavelet coefficients through the multi-scale wavelet transforms. However, with utilizing the different wavelet function, the corresponding results can be obtained. In this paper, we propose an adaptive referenced wave function to fit the periodic active components best in a least-squares sense. The results of experiment indicate our method has better validity and reliability.