Your browser doesn't support javascript.
loading
Posterior temperature optimized Bayesian models for inverse problems in medical imaging.
Laves, Max-Heinrich; Tölle, Malte; Schlaefer, Alexander; Engelhardt, Sandy.
Afiliación
  • Laves MH; Institute of Medical Technology and Intelligent Systems, Hamburg University of Technology, Am Schwarzenberg-Campus 3, Hamburg 21073, Germany. Electronic address: max-heinrich.laves@tuhh.de.
  • Tölle M; Group Artificial Intelligence in Cardiovascular Medicine, Heidelberg University Hospital, Im Neuenheimer Feld 410, Heidelberg 69120, Germany.
  • Schlaefer A; Institute of Medical Technology and Intelligent Systems, Hamburg University of Technology, Am Schwarzenberg-Campus 3, Hamburg 21073, Germany.
  • Engelhardt S; Group Artificial Intelligence in Cardiovascular Medicine, Heidelberg University Hospital, Im Neuenheimer Feld 410, Heidelberg 69120, Germany.
Med Image Anal ; 78: 102382, 2022 05.
Article en En | MEDLINE | ID: mdl-35183875
We present Posterior Temperature Optimized Bayesian Inverse Models (POTOBIM), an unsupervised Bayesian approach to inverse problems in medical imaging using mean-field variational inference with a fully tempered posterior. Bayesian methods exhibit useful properties for approaching inverse tasks, such as tomographic reconstruction or image denoising. A suitable prior distribution introduces regularization, which is needed to solve the ill-posed problem and reduces overfitting the data. In practice, however, this often results in a suboptimal posterior temperature, and the full potential of the Bayesian approach is not being exploited. In POTOBIM, we optimize both the parameters of the prior distribution and the posterior temperature with respect to reconstruction accuracy using Bayesian optimization with Gaussian process regression. Our method is extensively evaluated on four different inverse tasks on a variety of modalities with images from public data sets and we demonstrate that an optimized posterior temperature outperforms both non-Bayesian and Bayesian approaches without temperature optimization. The use of an optimized prior distribution and posterior temperature leads to improved accuracy and uncertainty estimation and we show that it is sufficient to find these hyperparameters per task domain. Well-tempered posteriors yield calibrated uncertainty, which increases the reliability in the predictions. Our source code is publicly available at github.com/Cardio-AI/mfvi-dip-mia.
Asunto(s)
Palabras clave

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Asunto principal: Diagnóstico por Imagen Tipo de estudio: Diagnostic_studies / Prognostic_studies Límite: Humans Idioma: En Revista: Med Image Anal Asunto de la revista: DIAGNOSTICO POR IMAGEM Año: 2022 Tipo del documento: Article Pais de publicación: Países Bajos

Texto completo: 1 Colección: 01-internacional Base de datos: MEDLINE Asunto principal: Diagnóstico por Imagen Tipo de estudio: Diagnostic_studies / Prognostic_studies Límite: Humans Idioma: En Revista: Med Image Anal Asunto de la revista: DIAGNOSTICO POR IMAGEM Año: 2022 Tipo del documento: Article Pais de publicación: Países Bajos