Study Of Quaternion Interpolation Method Application To Cardiac Dt-mri Data

Diffusion tensor imaging, it has been a rapid developing imaging technology. Due to technical constraints, obtain the diffusion tensor image as low resolution, so need to interpolation processing. Diffusion tensor data is different from scalar data, it has been exist on the manifold, scalar interpolation method hasn’t been suitable for diffusion tensor images. In the diffusion tensor image interpolation, the traditional quaternion method to solve the Riemann interpolation architecture tensor of anisotropic damage problem, but it exists anisotropy scores not drab. To solve this problem, we propose improved spectral quaternion method and dual quaternion method.Improved Spectral quaternion interpolation method based on Spectrum quaternion interpolation, it has been based on anisotropic tensor set thresholds, using different interpolation schemes amendments spectrum rights quaternion interpolation weights, application of new weighting interpolation obtained results. The method than the quaternion spectral method has stronger noise immunity, and reduced spectrum quaternion method of fractional anisotropy not monotonous of degree.Tensor dual quaternion interpolation method is a new tensor interpolation method, through the tensor parameters and direction mapped to dual quaternion, the dual quaternion domain interpolation. Finally, interpolation result of the conversion to a tensor data. The method solves the spectrum quaternion interpolation method has been appeared fractional anisotropy value not monotonous issue, and it is in the tensor interpolation direction also smoother.In the simulation data and cardiac DT-MRI data, we presented interpolation method was validated and compared with other popular interpolation methods, experimental results show that our method has a smaller error in FA and strongly noise immunity. In the tensor direction of interpolation, compared with the Riemann interpolation frame, quaternion interpolation has been reduced computational complexity and improved interpolated smooth.