De-noising of magnetic resonance images using independent component analysis

Digital MR Image processing often requires a prior application of filters to reduce the noise level of the image while preserving important details. This may improve the quality of digital MR images and contribute to an accurate diagnosis. De-noising methods based on linear filters cannot preserve image structures such as edges in the same way that methods based on nonlinear filters can do it. Recently, a nonlinear de-noising method based on ICA has been introduced [1,2] for natural and artificial images. The functioning of the ICA de-noising method depends on the statistics of the images. In this paper, we show that MRI has statistics appropriate for ICA de-noising. ICA transform is applied on MRI and its 12 independent tissue components are separated and then by observing statistical properties of each component suitable sparse coding shrinkage function is applied for de-noising of each component. We demonstrate experimentally that ICA de-noising is a suitable method to remove the noise of digitized MRI.