Functional exploration of the brain by fMRI - Theoretical approach of spectral analysis

Functional magnetic resonance imaging (fMRI) permits to obtain physiological information about MRI signal, which is modulated by electrical, biochemical, and physiological properties of the cerebral. tissue. It is possible to characterize the brain interactions from an fMRI signal. Particularly, the use of a spectral. analysis at a given frequency allows access to the time series chronology, which occurs within various activated areas of the brain. Thus, spectral. parameters such as coherency and phase shift may be calculated from presupposed stationary stochastic signals and of an estimate of the cross-spectral power density function. Coherency describes a correlation structure in frequency domain between signals and thus allows obtaining an accurate estimate of the phase retation (time delay), which connects the signals between them. We describe in the last part of the article a calculation method integrating spectral information obtained previously and which makes it possible to evaluate the intensity of the existing interaction between two distinct cerebral areas. (c) 2007 Elsevier Masson SAS.