Deep Learning for Magnetic Resonance Image Reconstruction and Analysis

Add to your list(s) Download to your calendar using vCal

• Ms Chen Qin (Doctoral Researcher, Department of Computing, Imperial College London, United Kingdom)
• Friday 09 August 2019, 11:30-12:30
• Computer Science, The Sloman Lounge (UG).

If you have a question about this talk, please contact Hector Basevi.

Host: Prof Aleš Leonardis

Abstract: Recent advances in deep learning have shown great potentials in improving the entire medical imaging pipeline and achieving human-level performance, from image acquisition and reconstruction to disease diagnosis. In this talk, we will fully leverage the power of deep learning techniques for medical imaging. Specifically, we will mainly focus on Magnetic Resonance (MR) image reconstruction and analysis. Firstly, we will introduce our recent study on dynamic MR image reconstruction from highly undersampled k-space data. A CRNN (convolutional recurrent neural network) model will be presented where it models the traditional iterative optimisation process in a learning setting and is able to exploit the spatio-temporal redundancies effectively and efficiently. As a complementary, a k-t NEXT (k-t Network with X-f Transform) method will be introduced in which we proposed to recover image signals by alternating the reconstruction process between x-f space and image space in an iterative fashion. Secondly, we will present some of our previous research on MR image analysis including both image segmentation and image registration. For image segmentation, an uResNet (u-shaped residual network) approach will be introduced to address the differential segmentation of white matter hyperintensities and stroke lesions on brain MRI . For image registration, I will go through our latest work, an unsupervised multi-modal deformable image registration (UMDIR) method, where we proposed to address the multi-modal registration problem by reducing it to a mono-modal one via disentangled representations. Finally, a joint framework which simultaneously predict cardiac motion estimation and segmentation will be presented, where we showed that the segmentation and registration tasks are beneficial from each other.

Website: https://scholar.google.com/citations?user=mTWrOqHOqjoC&hl=en

This talk is part of the Artificial Intelligence and Natural Computation seminars series.

This talk is included in these lists:

• Artificial Intelligence and Natural Computation seminars
• Computer Science Departmental Series
• Computer Science Distinguished Seminars
• Computer Science, The Sloman Lounge (UG)
• computer sience

Note that ex-directory lists are not shown.