We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

eess.IV

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Electrical Engineering and Systems Science > Image and Video Processing

Title: Unsupervised MRI Reconstruction with Generative Adversarial Networks

Abstract: Deep learning-based image reconstruction methods have achieved promising results across multiple MRI applications. However, most approaches require large-scale fully-sampled ground truth data for supervised training. Acquiring fully-sampled data is often either difficult or impossible, particularly for dynamic contrast enhancement (DCE), 3D cardiac cine, and 4D flow. We present a deep learning framework for MRI reconstruction without any fully-sampled data using generative adversarial networks. We test the proposed method in two scenarios: retrospectively undersampled fast spin echo knee exams and prospectively undersampled abdominal DCE. The method recovers more anatomical structure compared to conventional methods.
Subjects: Image and Video Processing (eess.IV); Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2008.13065 [eess.IV]
  (or arXiv:2008.13065v1 [eess.IV] for this version)

Submission history

From: Elizabeth Cole [view email]
[v1] Sat, 29 Aug 2020 22:00:49 GMT (1852kb)

Link back to: arXiv, form interface, contact.