A method for removing one or more artifacts from a time series of magnetic resonance (MR) images is provided. In one embodiment of the invention, an MR time series of data sets is acquired while changing the k-space that is sampled at each time frame. By using the acquired data, information is produced for a plurality of MR images at different time points, wherein the images contain desired components and one or more artifacts. Changing the k-space locations that are sampled from time frame to time frame makes the artifacts behave in a signature way through time, so that they can be identified as being artifacts and be removed through temporal analysis.