Maximum-Entropy Density Estimation for MRI Stochastic Surrogate Models
Stochastic spectral methods can generate accurate compact stochastic models for electromagnetic problems with material and geometric uncertainties. This letter presents an improved implementation of the maximum-entropy algorithm to compute the density function of an obtained generalized polynomial c...
| Main Authors: | , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | en_US |
| Published: |
Institute of Electrical and Electronics Engineers (IEEE)
2017
|
| Online Access: | http://hdl.handle.net/1721.1/108399 https://orcid.org/0000-0002-5880-3151 |
| Summary: | Stochastic spectral methods can generate accurate compact stochastic models for electromagnetic problems with material and geometric uncertainties. This letter presents an improved implementation of the maximum-entropy algorithm to compute the density function of an obtained generalized polynomial chaos expansion in magnetic resonance imaging (MRI) applications. Instead of using statistical moments, we show that the expectations of some orthonormal polynomials can be better constraints for the optimization flow. The proposed algorithm is coupled with a finite element-boundary element method (FEM-BEM) domain decomposition field solver to obtain a robust computational prototyping for MRI problems with low- and high-dimensional uncertainties. |
|---|