Inside bold - why neuronal activity modulates MR signal?
Although fMRI methods are well recognized as a powerful tool in neuroimaging, they still suffer from lack of repeatable quantitative measures of effect studied, especially in disease group studies, where pathologically changed brain morphologies or distortions caused by head movement arise. In such cases where typical statistical parametric mapping (SPM) approach based on general linear model (GLM) modeling fails, additional methods, giving complementary measures to standard statistical maps SPM may comprise an alternative approach. This subject covers two basic methodologies, which will be presented in this presentation: (1) A novel technique called Mean Regional Response (MRR) uses features derived from time-intensity curves in anatomically or functionally defi ned regions of interest (ROIs). Resulting features can be used for higher level analysis, like ANOVA or highlight differences between groups of subjects allowing inference about group separation; (2) Structural Equation Modeling (SEM) technique is focused on the extraction of connections between the active regions. Although SEM still do not allow arbitrary connection patterns, its use allows determination of time varying and bi-directional, asymmetric connections for groups of interest (patient vs. controls etc.). A methodological basics of BOLD modeling, which can be considered as a basis for most detection and classifi cation methods, will be presented and illustrated with author’s fMRI studies