Zero-filling Technique in Fresnel Transfrom Image Reconstruction for MR Image Denoising

Abstract

The Fresnel transform imaging method with a nonlinear field gradient for phase encoding has a great property that it can diffuse the noise contained in nuclear magnetic resonance (NMR) signal. But we cannot find any diffuseness of the noise, when we fix the signal matrix size to reconstruct images. However, if we prepare suitable area for the noise diffusing by zero-filling technique before reconstructing images in this method, the spectrum of noise will evenly distribute over all the zero filled area, on the other hand, we can obtain a multi-resolution type image, because the image frequency is split into equal intervals. After removing the noise by a threshold filter in the multi-resolution space, returning the denoised signal to beginning space by taking a series of inverseprocesses and reconstructing it with usual Fresnel transform imaging method, we can obtain an MR image that the noise is greatly improved, whereas the deterioration in spatial resolution is hardly caused.