Bayesian Optimization of Sampling Densities in MRI

Alban Gossard1,20000-0001-6782-5080, Frédéric de Gournay1,30000-0003-4721-3137, Pierre Weiss1,4,50000-0001-9083-214X
1: Institut de Mathématiques de Toulouse (IMT); UMR5219; Université de Toulouse; CNRS, France, 2: Université Paul Sabatier, F-31062 Toulouse Cedex 9, France, 3: INSA de Toulouse; F-31077 Toulouse, France, 4: Université de Toulouse, CNRS, 5: Centre de Biologie Intégrative (CBI), Laboratoire MCD, F-31062 Toulouse Cedex, France
Publication date: 2023/06/15
PDF · HAL · arXiv


Data-driven optimization of sampling patterns in MRI has recently received a significant attention. Following recent observations on the combinatorial number of minimizers in off-the-grid optimization, we propose a framework to globally optimize the sampling densities using Bayesian optimization. Using a dimension reduction technique, we optimize the sampling trajectories more than 20 times faster than conventional off-the-grid methods, with a restricted number of training samples. This method – among other benefits – discards the need of automatic differentiation. Its performance is slightly worse than state-of-the-art learned trajectories since it reduces the space of admissible trajectories, but comes with significant computational advantages. Other contributions include: i) a careful evaluation of the distance in probability space to generate trajectories ii) a specific training procedure on families of operators for unrolled reconstruction networks and iii) a gradient projection based scheme for trajectory optimization.


Sampling theory · compressed sensing · MRI · Fourier transform · data-driven optimization · globalization · Bayesian optimization

Bibtex @article{melba:2023:009:gossard, title = "Bayesian Optimization of Sampling Densities in MRI", author = "Gossard, Alban and de Gournay, Frédéric and Weiss, Pierre", journal = "Machine Learning for Biomedical Imaging", volume = "2", issue = "June 2023 issue", year = "2023", pages = "253--287", issn = "2766-905X", doi = "", url = "" }
RISTY - JOUR AU - Gossard, Alban AU - de Gournay, Frédéric AU - Weiss, Pierre PY - 2023 TI - Bayesian Optimization of Sampling Densities in MRI T2 - Machine Learning for Biomedical Imaging VL - 2 IS - June 2023 issue SP - 253 EP - 287 SN - 2766-905X DO - UR - ER -

2023:009 cover