Explore a 51-minute conference talk on deep learning-based MR image reconstruction and contrast conversion. Delve into advanced techniques for magnetic resonance imaging, including parallel imaging with deep learning and contrast conversion from multiple weighted images. Learn about applications of deep learning in k-space and image space, as well as the use of variational networks, deep cascade networks, and multi-stream CNNs. Discover innovative approaches such as domain transform learning and automated concepts in image reconstruction. Gain insights into parameter mapping, reconstruction frameworks, and their results. This comprehensive presentation by Dosik Hwang from Yonsei University, delivered at the Institute for Pure & Applied Mathematics at UCLA, offers a thorough exploration of cutting-edge developments in MR imaging technology.
Deep Learning-Based MR Image Reconstruction and Contrast Conversion
Institute for Pure & Applied Mathematics (IPAM) via YouTube
Overview
Syllabus
Introduction
Application of Deep Learning
Conventional Image Reconstruction
Domain
Method
Math
Variational Network
Deep Cascade Network
K Space
Dual Domain Approach
Intermediate Results
Domain Transform Learning
Automate Concept
Proposed Network
Results
Parallel Imaging
MultiStream CNN
Conclusion
Parameter Mapping
Reconstruction Framework
Reconstruction Results
Contrast Conversion
Summary
Taught by
Institute for Pure & Applied Mathematics (IPAM)
Related Courses
-
Introduction to Deep Learning
3.7 -
Deep Subspace Learning for Dynamic MR Image Reconstruction
-
Geometric Understanding of Supervised and Unsupervised Deep Learning for Biomedical Image Reconstruction
-
Deep-Learning Solutions for Efficient Cardiac MR
-
MR Image Acquisition and Reconstruction in the Era of Deep Learning
-
Deep Image Reconstruction from Human Brain Activity
