Bayesian non-linear regression with spatial priors for noise reduction and error estimation in quantitative MRI with an application in T1 estimation
| Autor: | Tommy Löfstedt, Anders Garpebring, Max Hellström, Mikael Bylund |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2020 |
| Předmět: |
Computer science
Maximum likelihood Posterior probability Bayesian probability Signal-To-Noise Ratio Bayesian statistics quantitative MRI 030218 nuclear medicine & medical imaging 03 medical and health sciences symbols.namesake 0302 clinical medicine Datorseende och robotik (autonoma system) Prior probability WAIC Radiology Nuclear Medicine and imaging Sannolikhetsteori och statistik Probability Theory and Statistics Computer Vision and Robotics (Autonomous Systems) Hyperparameter noise reduction Radiological and Ultrasound Technology Uncertainty Sampling (statistics) Bayes Theorem Markov chain Monte Carlo Regression analysis Image Enhancement Magnetic Resonance Imaging Markov Chains Nonlinear Dynamics 030220 oncology & carcinogenesis symbols tissue parameter estimation Radiologi och bildbehandling Monte Carlo Method Nonlinear regression Algorithm Algorithms Radiology Nuclear Medicine and Medical Imaging |
| Popis: | Purpose. To develop a method that can reduce and estimate uncertainty in quantitative MR parameter maps without the need for hand-tuning of any hyperparameters. Methods. We present an estimation method where uncertainties are reduced by incorporating information on spatial correlations between neighbouring voxels. The method is based on a Bayesian hierarchical non-linear regression model, where the parameters of interest are sampled, using Markov chain Monte Carlo (MCMC), from a high-dimensional posterior distribution with a spatial prior. The degree to which the prior affects the model is determined by an automatic hyperparameter search using an information criterion and is, therefore, free from manual user-dependent tuning. The samples obtained further provide a convenient means to obtain uncertainties in both voxels and regions. The developed method was evaluated on T 1 estimations based on the variable flip angle method. Results. The proposed method delivers noise-reduced T 1 parameter maps with associated error estimates by combining MCMC sampling, the widely applicable information criterion, and total variation-based denoising. The proposed method results in an overall decrease in estimation error when compared to conventional voxel-wise maximum likelihood estimation. However, this comes with an increased bias in some regions, predominately at tissue interfaces, as well as an increase in computational time. Conclusions. This study provides a method that generates more precise estimates compared to the conventional method, without incorporating user subjectivity, and with the added benefit of uncertainty estimation. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|