MRI imaging with noise filtering in wavelet space

A method of imaging of an object. A plurality of image datasets, for example a real part and an imaginary part of a magnetic resonance image, are acquired, and are transformed separately to a transform space and filtered. The filtered datasets are inverse transformed and recombined nonlinearly to produce the final image. The preferred transformation is a wavelet transformation. Noisy transform coefficients are suppressed by thresholding.