Variational manifold learning from incomplete data: application to multislice dynamic MRI

Qing Zou; Abdul Haseeb Ahmed; Prashant Nagpal; Sarv Priya; Rolf Schulte; Mathews Jacob

doi:10.48550/arXiv.2101.08196

Back

Variational manifold learning from incomplete data: application to multislice dynamic MRI

Preprint

Open access

Variational manifold learning from incomplete data: application to multislice dynamic MRI

Qing Zou, Abdul Haseeb Ahmed, Prashant Nagpal, Sarv Priya, Rolf Schulte and Mathews Jacob

ArXiv.org

03/20/2021

DOI: 10.48550/arXiv.2101.08196

Files and links (1)

url

https://arxiv.org/abs/2101.08196View

Preprint (Author's original)This preprint has not been evaluated by subject experts through peer review. Preprints may undergo extensive changes and/or become peer-reviewed journal articles. Open Access

Abstract

Current deep learning-based manifold learning algorithms such as the variational autoencoder (VAE) require fully sampled data to learn the probability density of real-world datasets. Once learned, the density can be used for a variety of tasks, including data imputation. However, fully sampled data is often unavailable in a variety of problems, including the recovery of dynamic and high-resolution MRI data considered in this work. To overcome this problem, we introduce a novel variational approach to learn a manifold from undersampled data. The VAE uses a decoder fed by latent vectors, drawn from a conditional density estimated from the fully sampled images using an encoder. Since fully sampled images are not available in our setting, we approximate the conditional density of the latent vectors by a parametric model whose parameters are estimated from the undersampled measurements using back-propagation. We use the framework for the joint alignment and recovery of multislice free breathing and ungated cardiac MRI data from highly undersampled measurements. Most of the current self-gating and manifold cardiac MRI approaches consider the independent recovery of images from each slice; these methods are not capable of exploiting the inter-slice redundancies in the datasets and require sophisticated post-processing or manual approaches to align the images from different slices. By contrast, the proposed scheme is able to align the multislice data and exploit the redundancies. Experimental results demonstrate the utility of the proposed scheme in dynamic imaging alignment and reconstructions.

Details

Title: Subtitle: Variational manifold learning from incomplete data: application to multislice dynamic MRI
Creators: Qing Zou
Abdul Haseeb Ahmed
Prashant Nagpal
Sarv Priya
Rolf Schulte
Mathews Jacob
Resource Type: Preprint
Publication Details: ArXiv.org
DOI: 10.48550/arXiv.2101.08196
ISSN: 2331-8422
Language: English
Date posted: 03/20/2021
Date updated: 12/10/2021
Academic Unit: Electrical and Computer Engineering; Iowa Neuroscience Institute; Iowa Technology Institute; Radiation Oncology; Radiology; Roy J. Carver Department of Biomedical Engineering
Record Identifier: 9984071601802771

Metrics

22 Record Views