Meeting Banner
Abstract #0621

In Vivo Human Spinal Cord Diffusion Tensor Imaging Using Rician Noise Filter

Dhanashree Vernekar1, Wenshu Qian2, Zhongping Zhang3, Pek-Lan Khong3, Mina Kim3

1Department of Diagnostic Radiology , The University of Hong Kong , Hong Kong, China; 2Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong , China; 3Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong, China

Diffusion tensor imaging (DTI) has been used to successfully show changes in white matter structure and connectivity in spinal cord of patients with various diseases. However, spinal cord DTI is still in its infancy due to technical challenges including intrinsically low signal-to-noise ratio. Here we propose to use a nonlocal means Rician noise filter to enhance the accuracy of tensor estimation and obtain robust DTI-derived measures. Our results show that Rician denoising can significantly decrease erroneous tensor estiamtions with reduced mean Chi-square up to 43% over without Rician denoising.

Keywords

ability accuracy accurately additional agree aimed analyses anisotropy applications approved automated available basis best bias board body capability cervical channel china clinical coil column columns combination commercially compute connectivity considered cord coupe covered custom decreased degree derived described diffusion diffusivity directed diseases dorsal drawn easy efficiency elsewhere erroneous error estimation evaluation excitation females filter fine fitting five fractional freedom full goodness health healthy hong human implement implementing improve improved in vivo induced informed institutes institutional investigating johns known lateral local location male manually maps measures metrics monitor national nature noise nominal nonlocal optimized parallel people portions preserved prior procedure properties quantify radiology random reception reduced registered registration republic resolution respectively rest review rigid role scanner sclerosis shot signed significantly similarly slice software spinal square statistical structure structures studies subject subjects subsequently successfully suffers suggesting supplied table tensor tissue towards transferred transformation unit upper useful various version volumes volunteers width woods written years yielding