Comparison of motion correction methods including particle filter for functional magnetic resonance imaging

Motion correction using particle filter is proposed and the comparison of three algorithms for human head motion correction of functional magnetic resonance imaging are performed. The traditional algorithm of gradient descent method, random sampling, and proposed particle filtering are used to correct the motion of real scanned human brain images. The result shows the particle filter achieves reasonably high speed and estimation of the correct position as well as the gradient descent method. The difference of estimation of these algorithms is small and the particle filtering may be better under more worst condition of head motion.