fMRI artifacts reduction using Bayesian image processing

Functional MRI is known to be prone to artifacts caused by spatio-temporally varying structural noise components such as gross head motion, CSF pulsation, physiological fluctuation, and magnetic susceptibility changes. The presence of these artifacts can cause negative and positive false activation and obscure true activated pixels. Thus the reliability of the functional images can be diminished. The application of image registration or noise filtering can reduce artifacts related to motion and physiological pulsation, but cannot correct the locally differing spatial variations in T2* signal loss and the image distortions due to off-resonance effects. In our work, Bayesian image processing is applied to reduce noise and artifacts and to enhance true activity detection. The results indicate that Bayesian processing is effective in reducing fMRI noise and artifacts and in improving true activity detection by enhancing the connectivity of the activated pixels.