ACMD Seminar: CLAIRE: Scalable multi-GPU algorithms for diffeomorphic image registration in 3D

Andreas Mang
Assistant Professor, Mathematics Dept., University of Houston

Thursday, March 23, 2023, 1:00-2:00 PM MT (3:00-4:00 PM ET)

A video of this talk will be made available to NIST staff in the Math channel on NISTube, which is accessible from the NIST internal home page. It will be taken down from NISTube after 12 months at which point it can be requested by emailing the ACMD Seminar Chair.

Abstract: In this talk, we will discuss (i) efficient numerical techniques for the solution of mixed-type PDE-constrained optimization problems with application to diffeomorphic image registration and (ii) the deployment of our algorithms to high-performance computing platforms. Our contributions are in the design of numerical methods and computational kernels that scale on heterogeneous, high-performance computing platforms.

Diffeomorphic registration is an infinite-dimensional, nonlinear inverse problem. The inputs are two views (images) of the same scene or object. Given these views, we seek an admissible spatial transformation y that relates points in one view to its corresponding points in the other. We formulate this problem as a constrained optimization problem with dynamical systems as constraints. We introduce a pseudo-time variable t and parameterize the sought-after mapping y by its velocity v. Prescribing suitable regularity requirements for v allows us to limit the space of admissible mappings y to R³-diffeomorphism. We will explore different formulations and discuss various numerical solution strategies.

Our solvers are based on state-of-the-art algorithms to enable fast convergence and short runtime. We will showcase results on real and synthetic data to study the rate of convergence, time-to-solution, numerical accuracy, and scalability of our solvers. As a highlight, we will showcase results for a GPU-accelerated implementation termed CLAIRE that allows us to solve clinically relevant 3D image registration problems to high accuracy in under 5 seconds on a single GPU, and scales up to 100s of GPUs.

Bio: Andreas Mang is an Assistant Professor for Applied Mathematics at the Department of Mathematics of the University of Houston. He received his PhD from the University of Luebeck (DE) in 2013. After that, he joined the Oden Institute for Computational Engineering and Sciences at The University of Texas at Austin (US) in 2013 for a postdoctoral fellowship. In 2017, he joined the Department of Mathematics at the University of Houston, where he currently holds a tenure-track Assistant Professorship for Applied Mathematics. His research interests include statistical and deterministic inverse problems, nonlinear optimal control, numerical optimization, data-enabled sciences, and parallel scientific computing. He works on the design, analysis, and deployment of effective numerical methods and parallel algorithms that deliver optimal performance and scale on high-performance computing platforms. In 2022, he received the NSF CAREER award.

Host: Gunay Dogan

Note: This talk will be recorded to provide access to NIST staff and associates who could not be present to the time of the seminar. The recording will be made available in the Math channel on NISTube, which is accessible only on the NIST internal network. This recording could be released to the public through a Freedom of Information Act (FOIA) request. Do not discuss or visually present any sensitive (CUI/PII/BII) material. Ensure that no inappropriate material or any minors are contained within the background of any recording. (To facilitate this, we request that cameras of attendees are muted except when asking questions.)

Note: Visitors from outside NIST must contact Lochi Orr (301) 975-3800; at least 24 hours in advance.