Abstract
We describe and evaluate a pre-processing method based on a periodic spiral sampling of diffusion-gradient directions for high angular resolution diffusion magnetic resonance imaging. Our pre-processing method incorporates prior knowledge about the acquired diffusion-weighted signal, facilitating noise reduction. Periodic spiral sampling of gradient direction encodings results in an acquired signal in each voxel that is pseudo-periodic with characteristics that allow separation of low-frequency signal from high frequency noise. Consequently, it enhances local reconstruction of the orientation distribution function used to define fiber tracks in the brain. Denoising with periodic spiral sampling was tested using synthetic data and in vivo human brain images. The level of improvement in signal-to-noise ratio and in the accuracy of local reconstruction of fiber tracks was significantly improved using our method.
Highlights
Diffusion-weighted imaging (DWI) is a powerful tool for inferring tissue structure and has been used extensively to map white matter pathways in healthy and diseased brains [1, 2]
PHANTOM DATA Signal-to-noise ratio Voxel-level mean squared error (MSE) comparison across different datasets showed that mean MSE is significantly decreased when lop-DWI is lop-DWI: a pre-processing method
Under-estimation rate of lop-Q-Ball Imaging (QBI) were slightly higher than QBI, and reached similar value as signal-to-noise ratio (SNR) increased
Summary
Diffusion-weighted imaging (DWI) is a powerful tool for inferring tissue structure and has been used extensively to map white matter pathways in healthy and diseased brains [1, 2]. Errors due to thermal and physiological noise and to eddy currents affect individual diffusion-weighted images and influence the accuracy of measures obtained from DWI data such as fractional anisotropy, fiber orientation, and separation between fibers [3,4,5]. Parametric and non-parametric statistical approaches have been used to describe the noise distribution and to derive the best fit for tensor parameters [6,7,8]. Parametric methods rely on the distributional model of noise being correct. Non-parametric statistical approaches are model independent but are less powerful in hypothesis testing and more computationally demanding [7]
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
