http://ajer.org/papers/v5(02)/F0502042048.pdf4D Collaborative Non-Local Means Based Diffusion MRI Denoising
Journal Title: American journal of Engineering Research - Year 2016, Vol 5, Issue 2
Abstract
Noise is a major issue that reduces the quality of images acquired by diffusion MRI (dMRI). Recently, the non-local means (NLM) algorithm has been proposed and successfully applied in dMRI denoising. However, NLM relies on self-similarity information and tends to fails when recurrent image structures cannot be located. To address this issue, we introduce the improved collaborative NLM. Both inner-image and interimage similarity information are used. Specifically, a group of co-denoising images are first registered to the target space. NLM-like block matching is then performed on both target noisy image and co-denoising images. This formulation can significantly increase the amount of similarity information and reduce the rare patch effect. Moreover, in order to adapt to the characteristics of dMRI, we present a complete denoising framework with multiple techniques including 4D image block, pseudo-residual-based noise standard deviation estimation, Rician bias correction, and block preselection. Extensive experiments on both synthetic and real data demonstrate that the proposed framework outperforms the classical NLM method. Keywords: Block Matching, Denoising, Diffusion MRI, Non-Local Means, Kernel Regression
Authors and Affiliations
Geng Chen,, Yafeng Wu*
Keywords
Related Articles
- EP ID EP402480
- DOI -
- Views 38
- Downloads 0
Hamilton Mechanical Equations with Four Almost Complex Structures on Riemannian Geometry
The study concerns the Hamilton Mechanical Equations with Four Almost Complex Structures on Riemannian geometry. The four complex structures have been derived also important applications of Hamiltonian mechanical systems...
Geotechnical Properties of Lateritic Overburden Materials on the Charnockite and Gneiss Complexes in Ipele-Owo Area, Southwestern Nigeria
The lateritic soils that overburden Charnockite and Gneiss Complexes in Ipele-Owo area of Ondo State, Nigeria are in great use for local construction. However, little is known of their engineering properties and suitabil...
Filtered Orthogonal Frequency Division Multiplexing: A Waveform Candidate for 5G
The emerging Internet of Things will make the next generation 5G systems to support a broad range of diverse needs with greater efficiency requirements. The new class of services will need a higher data rates, to handle...
Development of Decision Support System for Vendor Selection Using Ahp-Vikor Based Hybrid Approach
Supply chain management (SCM) has emerged as an increasingly important approach to improve the performance of logistics systems. SCM is an integrated approach to increase the effectiveness of the logistics chain by impro...
A Survey Paper on Real Time Detection & Reporting By Social Networks
In today’s era social network like Twitter, Facebook, LinkedIn playing very vital role in human behavior & their day to day life. These social networking sites can be used as social sensors such that we can predict or re...