Current browse context:
eess.IV
Change to browse by:
References & Citations
Electrical Engineering and Systems Science > Image and Video Processing
Title: Preconditioned P-ULA for Joint Deconvolution-Segmentation of Ultrasound Images -- Extended Version
(Submitted on 19 Mar 2019 (v1), last revised 21 Jan 2020 (this version, v4))
Abstract: Joint deconvolution and segmentation of ultrasound images is a challenging problem in medical imaging. By adopting a hierarchical Bayesian model, we propose an accelerated Markov chain Monte Carlo scheme where the tissue reflectivity function is sampled thanks to a recently introduced proximal unadjusted Langevin algorithm. This new approach is combined with a forward-backward step and a preconditioning strategy to accelerate the convergence, and with a method based on the majorization-minimization principle to solve the inner nonconvex minimization problems. As demonstrated in numerical experiments conducted on both simulated and in vivo ultrasound images, the proposed method provides high-quality restoration and segmentation results and is up to six times faster than an existing Hamiltonian Monte Carlo method.
Submission history
From: Marie-Caroline Corbineau [view email][v1] Tue, 19 Mar 2019 16:59:32 GMT (2690kb,D)
[v2] Fri, 14 Jun 2019 09:57:49 GMT (3560kb,D)
[v3] Thu, 1 Aug 2019 16:07:06 GMT (3441kb,D)
[v4] Tue, 21 Jan 2020 19:04:12 GMT (3432kb,D)
Link back to: arXiv, form interface, contact.