A review on noise estimation methods in multi-coil MRI

Noise in an image is variation of information from the actual data. If there is any noise in the medical image then consequence will be huge. Physician won't be able to do proper diagnose of the disease and can also affects the quality of post processing techniques like registration and segmentation. Modern techniques use multiple coil MRI(Magnetic Resonance Imaging) where noise varies with position in the image i.e., non-stationary noise. Many methods are available for the non-stationary noise estimation in the literature but those needs multiple acquisitions and additional details. These limitations are avoided in homomorphic approach. Aja-Fernandez et.al used homomorphic approach on Devore et al. and is found to be a good method. In this paper homomorphic approach is applied on different known noise estimation methods in multi-coil MRI and a comparison is made. Visual comparison and quantitative analysis shows that homomorphic approach on Devore et.al gives better result for both Gausssian and Rician than other methods.