Robust Anisotropic Diffusion to Produce Clear Statistical Parametric Map from Noisy fMRI

Functional magnetic resonance imaging (fMRI) uses MRI to noninvasively map areas of increased neuronal activity in human brain without the use of an exogenous contrast agent. Low signal-to-noise ratio of fMRI images makes it necessary to use sophisticated image processing techniques, such as statistical parametric map (SPM), to detect activated brain areas. This paper presents a new technique to obtain clear SPM from noisy fMRI data. It is based on the robust anisotropic diffusion. A direct application of the anisotropic diffusion to fMRI does not work, mainly due to the lack of sharp boundaries between activated and non-activated regions. To overcome this difficulty, we propose to calculate SPM from noisy fMRI, compute diffusion coefficients in the SPM space, and then perform the diffusion in fMRI images using the coefficients previously computed. These steps are iterated until the convergence. Experimental results using the new technique yielded surprisingly sharp and noiseless SPMs.