This paper presents techniques for analyzing and visualizing dynamic brain imaging data, including functional magnetic resonance imaging (fMRI), optical imaging, and magnetoencephalography (MEG). The methods focus on multivariate time series analysis, particularly using multitaper spectral analysis to separate signal from noise and characterize the signal. The techniques involve two stages: noise characterization and suppression, and signal characterization and visualization. A key contribution is the development of a space-frequency singular value decomposition (SVD) technique for characterizing image data and an algorithm for removing physiological artifacts. The paper also discusses the challenges of analyzing multivariate time series data, including non-stationarity and the need for frequency-based representations with short, moving windows. It reviews different brain imaging techniques, their spatiotemporal resolutions, and sources of noise. The paper also describes data sets used for analysis and presents methods for spectral analysis, including multitaper spectral estimation, autoregressive spectral estimation, and conventional spectral analysis. The paper emphasizes the importance of spectral analysis in understanding the dynamics of neural systems and the need for appropriate analytical tools to extract meaningful information from complex brain imaging data.This paper presents techniques for analyzing and visualizing dynamic brain imaging data, including functional magnetic resonance imaging (fMRI), optical imaging, and magnetoencephalography (MEG). The methods focus on multivariate time series analysis, particularly using multitaper spectral analysis to separate signal from noise and characterize the signal. The techniques involve two stages: noise characterization and suppression, and signal characterization and visualization. A key contribution is the development of a space-frequency singular value decomposition (SVD) technique for characterizing image data and an algorithm for removing physiological artifacts. The paper also discusses the challenges of analyzing multivariate time series data, including non-stationarity and the need for frequency-based representations with short, moving windows. It reviews different brain imaging techniques, their spatiotemporal resolutions, and sources of noise. The paper also describes data sets used for analysis and presents methods for spectral analysis, including multitaper spectral estimation, autoregressive spectral estimation, and conventional spectral analysis. The paper emphasizes the importance of spectral analysis in understanding the dynamics of neural systems and the need for appropriate analytical tools to extract meaningful information from complex brain imaging data.