Summary

Computer vision and image analysis are both great examples of successes with machine learning especially deep learning. While computer vision and image analysis primarily deal with existing images, tomographic reconstruction produces images of internal structures from externally measured indirect data. Recently, deep learning techniques are being actively explored for tomographic reconstruction by multiple groups worldwide, with encouraging results and potential major impacts. We believe that deep reconstruction is a next major target of deep learning. Sponsored by GE Research, we are honored to host this workshop for brainstorming and collaboration.

Agenda

Time Topic Speaker
SESSION 1
8:00AM Opening remarks
8:10AM Deep MR Reconstruction: Towards Plug & Play MRI (video available) Chris Hardy, GRC
8:40 AM Deep Learning-based Universal Beamforming for Ultrasound Imaging (video and pdf available) Jong Chul Ye, KAIST
9:10 AM Deep Learning in CT Image Reconstruction: Sinogram In, Image Out via iCT-Net (video available) Guang-Hong Chen, UW Madison
9:40 AM Deep Learning Based Iterative PET Image Recon: Populational vs Personalize (video and pdf available) Quanzheng Li, MGH
10:10 AM Coffee break
SESSION 2
10:30 AM Speeding up MRI with deep learning: Denoising fewer averages & reconstructing fewer samples Ricardo Otazo, MSK
11:00 AM Principle-task-driven Machine Learning for Low-dose Computed Tomography Reconstruction (video and ppt available) Jerome Liang, Stony Brook U
11:30 AM Cutting through the fog: Removing clutter in ultrasound B-Mode (video available) Brett Byram, Vanderbilt U
12:00 PM Lunch
SESSION 3
1:00 PM Life at the bottom: how low SNR inspired a new reconstruction paradigm (video available) Matthew Rosen, MGH
1:30 PM Machine learning in spectral CT(video, pdf and ppt available) Hengyong Yu, U Mass-Lowell
2:00 PM Deep learning in quantitative SPECT and PET image reconstruction and processing (video and pdf available) Chi Liu, Yale U
2:30 PM Ultrasound Imaging: Deep and Hybrid (video, pdf and ppt available) Ge Wang, RPI
3:00 PM Coffee break
SESSION 4
3:20 PM Reimagining MRI in the era of learned image reconstruction Patricia Johnson, NYU
3:50 PM A hierarchical approach to deep learning and its application to tomographic reconstruction (video available) Lin Fu, GRC
4:20 PM NIBIB perspective on Deep Imaging Behrouz Shabestari, NIBIB
4:50 PM Closing remarks
5:00 PM ADJOURN