Reducing chemical shift artifacts in MRI in time-varying gradients

An image processing algorithm to reduce chemical shift artifacts in MRI (magnetic resonance imaging) with time-varying gradients and to obtain the chemical shift images without increasing the imaging time is presented. It is known that when both gradients in two directions of the imaging plane are time-varying, such as in the spiral-type scanning MRI, the chemical shift artifacts appear as a blurred image. The artifacts cannot be neglected, since the total power of the out-of-focus point spread function (PSF) for the blurred image equals that of the in-focus PSF. The proposed algorithm separates the chemical shift images by iteratively estimating the artifacts under an a priori constraint that the original images should have a real and nonnegative value. Computer simulation results show that the average amplitude of the artifacts can be reduced significantly after several iterations, and that the algorithm works well even when the observation noise exists.