Curve and surface fitting with splines
Curve and surface fitting with splines
Variational Frameworks for DT-MRI Estimation, Regularization and Visualization
ICCV '03 Proceedings of the Ninth IEEE International Conference on Computer Vision - Volume 2
Fast and compact smoothing on large multidimensional grids
Computational Statistics & Data Analysis
Generalized structured additive regression based on Bayesian P-splines
Computational Statistics & Data Analysis
Hi-index | 0.03 |
The present methodological development and the primary application field originate from diffusion tensor imaging (DTI), a powerful nuclear magnetic resonance technique which enables the quantification of microscopical tissue properties. The current analysis framework of separate voxelwise regressions is reformulated as a 3D space-varying coefficient model (SVCM) for the entire set of diffusion tensor images recorded on a 3D voxel grid. The SVCM unifies the three-step cascade of standard data processing (voxelwise regression, smoothing, interpolation) into one framework based on B-spline basis functions. Thereby strength is borrowed from spatially correlated voxels to gain a regularization effect right at the estimation stage. Two SVCM variants are conceptualized: a full tensor product approach and a sequential approximation, rendering the SVCM numerically and computationally feasible even for the huge dimension of the joint model in a realistic setup. A simulation study shows that both approaches outperform the standard method of voxelwise regression with subsequent regularization. Application of the fast sequential method to real DTI data demonstrates the inherent ability to increase the grid resolution by evaluating the incorporated basis functions at intermediate points. The resulting continuous regularized tensor field may serve as basis for multiple applications, yet, ameloriation of local adaptivity is desirable.