Probabilistic Analysis of Functional Magnetic Resonance Imaging Data英文文献.pdfVIP

Probabilistic Analysis of Functional Magnetic Resonance Imaging Data Lawrence R. Frank, Richard B. Buxton, Eric C. Wong Probability theory is applied to the analysis of fMRl data. The tions induced by well known stimulations. In this paper posterior distributionof the parameters is shown to incorpo- we will analyze this simplest case with a general ap- rate all the information availablefrom the data, the hypothe- proach to the problem that allows the estimation of both ses, and the prior information. Under appropriate simplifying model parameters and our certainty in these estimates, in conditions,the theory reducesto the standard statistical test, a way flexible enough to incorporate the available prior including the general linear model. The theory is particularly information while retaining logical consistency. suited to handle the spatial variations in the noise present in fMRI, allowing the comparison of activated voxels that have We present a general probabilistic method in the style different, and unknown, noise. The theory also explicitly in- of Jaynes (9-11)for the analysis of fMRI data, which cludes prior information,which is shown to be critical in the allows estimation of the parameters of a given model, the attainment of reliable activation maps. incorporation of relevant prior information, estimation of Key words: fMRl data analysis; probability theory;time series the noise, and comparison of data. Under appropriate analysis. conditions this method reproduces the results of stan- dard statistical analyses, such as the general linear