Plug-and-Play Priors for Reconstruction-Based Placental Image Registration

Abstract

This paper presents a novel deformable registration framework, leveraging an image prior specified through a denoising function, for severely noise-corrupted placental images. Recent work on plug-andplay (PnP) priors has shown the state-of-the-art performance of reconstruction algorithms under such priors in a range of imaging applications. Integration of powerful image denoisers into advanced registration methods provides our model with a flexibility to accommodate datasets that have low signal-to-noise ratios (SNRs). We demonstrate the performance of our method under a wide variety of denoising models in the context of diffeomorphic image registration. Experimental results show that our model substantially improves the accuracy of spatial alignment in applications of 3D in-utero diffusion-weighted MR images (DW-MRI) that suffer from low SNR and large spatial transformations.

Deep Learning for Regularization Prediction in Diffeomorphic Image Registration

“…Diffeomorphic image registration is a fundamental tool for various medical image analysis tasks, as it provides smooth and invertible smooth mapping (also known as a diffeomorphism) between pairwise images. Examples include atlas-based image segmentation (Ashburner and Friston, 2005;Gao et al, 2016), anatomical shape analysis based on geometric changes (Vaillant et al, 2004;Hong et al, 2017), and motion correction in spatial-temporal image sequences (De Craene et al, 2012;Liao et al, 2016;Xing et al, 2019). The nice properties of diffeomorphisms keep topological structures of objects intact in images.…”

Deep Learning for Regularization Prediction in Diffeomorphic Image Registration

“…Diffeomorphic image registration is a fundamental tool for various medical image analysis tasks, as it provides smooth and invertible smooth mapping (also known as a diffeomorphism) between pairwise images. Examples include atlas-based image segmentation (Ashburner and Friston, 2005;Gao et al, 2016), anatomical shape analysis based on geometric changes (Vaillant et al, 2004;Hong et al, 2017), and motion correction in spatial-temporal image sequences (De Craene et al, 2012;Liao et al, 2016;Xing et al, 2019). The nice properties of diffeomorphisms keep topological structures of objects intact in images.…”

Regularization by Denoising via Fixed-Point Projection (RED-PRO)

“…The PnP approach replaces the proximal operators with state-of-the-art denoisers, thus, embodying implicit priors for regularizing inverse problems. This framework has gained great interest due to its success in various applications [70,81,63,43,88,1,82], and it has been extended to other proximal algorithms such as proximal gradient method (PGM) [8,59], approximate message passing (AMP) [58,33,6] and half quadratic splitting [86]. Schemes similar to PnP has been proposed in [28] and [34], where the former is based on the augmented Lagrangian method and the latter relies on the notion of Nash equilibrium.…”