CIG is looking for motivated WashU students for the following projects. We have a history of mentoring undergraduate and graduate students interested in research, with many of our alumni becoming PhD students in top programs (including WashU) and many projects leading to publications in top venues (see below). If you are interested in a given project, please apply by emailing your resume and a short description of your interests to the project supervisor.

Deep learning for diagnosis of post-traumatic elbow disease

Status: reserved Description: Deep learning may revolutionize the study of musculoskeletal conditions by automating the evaluation pipeline. This project seeks to develop tissue and modality-specific DL algorithms for automatic evaluation of disease stage and explore multi-modal in post-traumatic elbow conditions using histology, computed tomography, and magnetic resonance imaging. This project is in collaboration with the Musculoskeletal Soft Tissue Laboratory. Supervision: Weijie Gan (weijie.gan@wustl.edu), Michael David, Spencer Lake, Ulugbek Kamilov Areas: Image Analysis, Deep Learning

Detecting multiple myeloma using deep image analysis

Status: available Description: Multiple myeloma (MM) is a devastating form of cancer. The development of imaging tools for MM is an active research area with the potential to enable breakthroughs in its diagnosis and treatment. CIG is looking for a motivated student to develop new deep learning methods for MM detection in the context of our collaboration with Prof. Monica Shokeen. Supervision: Weijie Gan (weijie.gan@wustl.edu), Monica Shokeen, Ulugbek Kamilov Areas: Image Analysis, Deep Learning

Inhomogeneity correction in quantitative brain MRI

Status: available Description: Magnetic field inhomogeneities adversely affect MRI images. The correction of such inhomogeneities can be formulated as an optimization problem that exploits their mathematical model and machine learning (ML). CIG is seeking a motivated student interested in using optimization and ML for medical imaging. Supervision: Weijie Gan (weijie.gan@wustl.edu), Xiaojian Xu, Dmitriy Yablonskiy, Ulugbek Kamilov Areas: Image Analysis, Deep Learning

Deep learning for quantitative brain imaging

Status: active (Eddie Chandler, Flora Sun) Description: Quantitative brain imaging is extensively used for early detection of neurodegenerative diseases such as Alzheimer disease and Dementia. This project seeks to build on the latest progress in self-supervised deep learning to enable the next generation algorithms that are robust to noise and other artifacts. This project is well-suited for those students interested in learning advanced concepts in deep learning and imaging. Supervision: Xiaojian Xu (xiaojianxu@wustl.edu), Ulugbek Kamilov Areas: Medical Imaging, Deep Learning

Deep learning for diagnosis of post-traumatic elbow disease

Status: active (Yanpeng Yuan) Description: Deep learning may revolutionize the study of musculoskeletal conditions by automating the evaluation pipeline. This project seeks to develop tissue and modality-specific DL algorithms for automatic evaluation of disease stage and explore multi-modal in post-traumatic elbow conditions using histology, computed tomography, and magnetic resonance imaging. This project is in collaboration with the Musculoskeletal Soft Tissue Laboratory. Supervision: Weijie Gan (weijie.gan@wustl.edu), Michael David, Spencer Lake, Ulugbek Kamilov Areas: Image Analysis, Deep Learning

Low-complexity learning for large-scale imaging

Status: active (Ziwen Wang) Description: Model-based deep learning refers to a new family of methods that integrate both physics-based and trainable parameters within a single neural network. For many large-scale applications, computing gradients of such networks via backpropagation is infeasible due to the memory limitations of GPUs. This project seeks to address this issue by designing memory-efficient architectures obtained by trading off computation storage and computational complexity. We will consider large-scale biomedical imaging applications. Supervision: Jiaming Liu (jiaming.liu@wustl.edu), Ulugbek Kamilov Areas: Deep Learning, Algorithm Design

Tunable deep learning for computational imaging

Status: active (Ann Zhou) Description: Deep learning has led to a paradigm shift in computational imaging. In this project, we will explore the topic of tunable deep learning, where the goal is to design techniques that enable us to adjust convolutional neural networks without retraining them. We will explore the applications of the techniques in biomedical imaging. Supervision: Xiaojian Xu (xiaojianxu@wustl.edu), Ulugbek Kamilov Areas: Deep Learning, Image Restoration

Synthesizing high-resolution videos using deep learning

Status: active (Wentao Shangguan) Description: Video reconstruction is aimed to restore clear videos from corrupted observations, where state-of-the-art methods have achieved promising performance via supervised deep learning technologies. However, those methods require ground-truth videos, which may be inaccessible in some cases. Our goal is to develop a novel video reconstruction algorithm without the need of ground-truth. Supervision: Yu Sun (sun.yu@wustl.edu), Ulugbek Kamilov Areas: Deep Learning, Low-Level and Physics-Based Vision

Accurate calibration of coil sensitivities in MRI

Status: active (Yuyang Hu) Description: Parallel MR image reconstruction yields optimal results when the coil sensitivity maps are accurately known. Unfortunately, in practical scenarios, obtaining accurate estimates of the sensitivity maps is not possible. In this project, we explore the potential of deep learning to enable accurate estimation of coil sensitivity maps. Supervision: Weijie Gan (weijie.gan@wustl.edu), Ulugbek Kamilov Areas: Deep Learning, Image Restoration

Next generation robust learning algorithms

Status: active (Renhao Liu) Description: Imposing mathematical properties to convolutional neural networks is an active area of research. In this project, we will specifically focus on training CNNs that are Lipschitz continuous. The application of such neural nets include computational imaging, generative models, and robustness in machine learning. Supervision: Yu Sun (sun.yu@wustl.edu), Ulugbek Kamilov Areas: Deep Learning, Large-Scale Optimization

• Deep Learning for quantitative brain imaging (E. Chandler, X. Xu, Y. Sun, and U. S. Kamilov)
• Cycle GAN for quantitative MRI (Y. Li, X. Xu, and U. S. Kamilov)
• Training robust convolutional neural networks (R. Liu, Y. Sun, and U. S. Kamilov)
• Implementing CNNs with Tensorflow 2.0 (Z. Wang, Y. Sun, and U. S. Kamilov)
• Autocalibration for regularization by denoising (M. Xie, Y. Sun, and U. S. Kamilov)

• Machine learning for 4D MRI (W. Gan, H. An, U. S. Kamilov)
• Proximal Optimization for Computerized Tomography (R. Schurz, J. Liu, U. S. Kamilov)
• End-to-End Training of VAE-GAN Priors (G. Meng, J. Liu, U. S. Kamilov)
• Design of GAN Prios for Iterative Algorithms (A. Bani, J. Liu, U. S. Kamilov)
• Recovery Experiments for Deep Learning (J. Liu, Y. Sun, U. S. Kamilov)
• Autocalibration for Regularization by Denoising (M. Xie, Y. Sun, and U. S. Kamilov)
• Stochastic Algorithms using Deep Priors (Z. Wu, Y. Sun, U. S. Kamilov)
• Fast Automatic Focal Plane Selection for Brightfield Microscopy using Machine Learning (Y. Song, X. Xu, N. Laird, Z. Pincus, U. S. Kamilov)
• Nonlocal Learning for Image Restoration (Y. Li, X. Xu, U. S. Kamilov)
• Motion-artifact Correction in Quantitative MRI (H. Tang, X. Xu, U. S. Kamilov)

• Motion artifact reduction in quantitative brain MRI (N. Woythaler, X. Xu, U. S. Kamilov)
• Large-Scale Optimization for Computational Imaging (Y. Jia, U. S. Kamilov)
• Generative Adversarial Networks for Image Restoration (R. Suzuki, X. Xu, U. S. Kamilov)
• Adversarially robust classifiers for image reconstruction (E. Tang, X. Xu, U. S. Kamilov)
• Stochastic Algorithms using Deep Priors (Z. Wu, Y. Sun, U. S. Kamilov)
• Asynchronous large-scale optimization for imaging (Y. Sun, Y. Sun, U. S. Kamilov)
• Next generation Cryo-EM imaging (M. Xie, Y. Sun, U. S. Kamilov)
• Computational methods for quantitative brain MR imaging (M. Torop, Y. Sun, U. S. Kamilov)

• Deep Single-Molecule Super-Resolution Imaging (S. Xu, Y. Sun, H. Mazidi, M. Lew, U. S. Kamilov)
• Learning for material imaging with light (Z. Wu, Y. Sun, U. S. Kamilov)
• Regularization with ConvNets (J. Liu, Y. Sun, U. S. Kamilov)

• Regularization with ConvNets (J. Liu, Y. Sun, U. S. Kamilov)
• Learning for 3D computational imaging (W. Gan, X. Xu, U. S. Kamilov)
• Convolutional dictionary learning for image reconstruction (F. Long, X. Xu, U. S. Kamilov)
• Unsupervised deep image learning (H. Zou, Y. Sun, U. S. Kamilov)
• Distributed algorithms for nonconvex stochastic optimization (M. Abijaoude, U. S. Kamilov)
• Image restoration with GANs (F. Zhai, Y. Sun, U. S. Kamilov)
• Learning for MRI (J. Xing, X. Xu, U. S. Kamilov)
• Learning for material imaging with light (Z. Wu, Y. Sun, U. S. Kamilov)
• Dictionary Learning for Fourier Ptychographic Microscopy (S. Xu, X. Xu, U. S. Kamilov)