Algorithm Research On MRI Denoising And Segmentation

This paper studies MRI denoising and segmentation. The diagnostic and visual quality of MRI are affected by noise of diverse origins while acquisition, mainly including noise from hardware and human factors. Noise reduced the quality of MRI, impact identification and analysis, so denoising is a key preprocessing step especially for noisy, low SNR MRI. Segmentation refers to the characteristics of image into regional extracts interested target technology and process. The research of MRI segmentation mainly concentrated on brain tumor segmentation, liver tumor segmentation, liver, lung, cardiac segmentation and so on. MRI segmentation is the foundation of quantitative analysis,3D reconstruction, auxiliary medical diagnosis. We propose two MRI denoising methods, apply improved level-set method to liver tumor segmentation, experimental results show desirable performances of our methods. The research works in this paper include:An efficient MRI denoising algorithm is designed to remove Rician noise. In the complex wavelet domain through DT-CWT, combined with BF, NeighShrinkSURE and BivariateShrink, an efficient MRI denoising algorithm, which fully considers the noise characteristics and wavelet coefficient's inter and intra-scale dependencies, is proposed. The performance of the proposed algorithm mainly depends on the estimation precision of the noise standard deviation in the coefficients of square MRI after DT-CWT transform, then relate to the parameters of BF and two shrink methods" weight factors. In order to make the cooperative method show the best performance, this paper takes mean square error (MSE), peak signal-to-noise ratio(PSNR), structural similarity index (SSIM) as the image quality evaluation indexes to correct traditional noise standard deviation estimation method, determine the parameters in BF and the weight factor between two shrink methods. This paper designs an efficient algorithm that combines three methods in the dual tree complex wavelet domain. Denoising through DT-CWT is superior to the basic wavelet transform, the filtering accounts for inter-scale dependency and neighboring similarities, the use of BF enhances the low frequency part of image, aiming at removing Rician noise in MRI. the proposed algorithm has better noise reduction while preserving image's edge and detail well.Allowing for the characteristic of squared-magnitude MR images which follows a non-central chi-square distribution, we use the Chi-Square Unbiased Risk Estimate (CURE) to determine an optimal threshold for NeighShrink. NeighShrinkCURE denoising algorithm is proposed. The use of bilateral filter and cycle spinning further enhance the denoising performance. The experimental results show that the proposed algorithm is simple and efficient, The comprehensive performance of the proposed algorithm is superior to several current MRI denoising algorithms in wavelet domain.We present a stochastic active contour model for liver tumor segmentation in MRI/CT. This model make use of region-based information with probability distribution characteristics and shape priors. Since medical images often has low contrast and blurred region boundaries, the segmentation process may easily cause boundary leakage. In order to avoid boundary leaking, likelihood estimation function and ellipse distance function are introduced to set up potential energy, the iterative formula is found by minimizes the energy functional, the assessment demonstrates good segmentation performance, avoid boundary leakage well.