John Hoffman
[email protected]
Hello!
This website is largely an effort to archive and keep a record of my work, as well as a sort of business card for my experience. Please feel free to reach out if you'd like to collaborate or hire me for consulting work.
JMH - December 28, 2023
Websites
Research Interests
I'm an imaging and computer vision scientist currently working in the UCLA Department of Radiology! Much of my work is predicated on the idea that a CT scan is more than just a single reconstruction, and that with access to the raw projection data we can learn substantially more from CT imaging than we currently do. By using first principles and physics, we can begin to understand how information, rather than just pixel values, flows through the mathematical reconstruction operation, and how that determines the inherent reliability of a scan, particularly for use in quantitative imaging. We hope that by understanding this flow of information, we can predict which imaging features will be more or less reliable for clinical diagnostic purposes.
We (as many others) are also looking to understand the coming roles that AI and machine learning can and will play in the days ahead, and also how we can improve AI performance to make it more suitable for clinical usage. In particular, we have begun investigating a "recon-in-the-loop" data augmentation strategy for AI, as well as some early investigations into automated phantom analysis trained using only synthetic models of the phantom; this continues into some explorations into manufacturing our own phantoms. We continue to expand this area of our research.
Computing
I personally am deeply interested in the computational details of all of the problems I work on and encourage my students to also dive deep to better understand these details. Computation lies at the core of modern medicine in every way (PACS, image viewers, scheduling, drug delivery, security of a hospital, CT image reconstruction, CT image acquisition, etc.) and impact not only the research that we do, but the real viability of certain techniques both for research and clinic. As with all science, through a better understanding our tools we gain access to new knowledge. My work has included GPU algorithm implementation with OpenCL and CUDA, CPU architecture-specific optimization, deployment of algorithms as services (server/backend deployment), low-level network programming for distributed computing, cloud development and deployment, and many other topics. I love seeing projects move from conception to deployment and start being used by real people. The hacker mentality of "crack it open and figure out how it works" runs deep.
Academics
- Ph.D. Biomedical Physics, UCLA, Los Angeles, CA (dissertation)
- B.S. Physics, Virginia Tech, Blacksburg, VA
- B.S. Mathematics, Virginia Tech, Blacksburg, VA
Job Experience
- Assistant Adjunct Professor, UCLA Radiological Sciences. May 2022, Present
- Lead Software Development Engineer, Magic Leap. April 2022, February 2022
- Computer Vision Engineer, 3D Reconstruction, Magic Leap. December 2018, April 2021
- Independent Contractor, iTomography. December 2017, May 2019
- Imaging Scientist, Toshiba (now Canon) Medical Research USA, Inc.. October 2016, December 2017
Selected Publications
- (link)(pdf) Scott Hsieh, John Hoffman, Frédéric Noo, Accelerating Iterative Coordinate Descent Using a Stored System Matrix. Medical Physics 46(12), e801-e809. (2019)
- (link)(pdf) John Hoffman, Nastaran Emaminejad, M. Wasil Wahi-Anwar, Grace Hyun J. Kim, Matthew Brown, Stefano Young, Michael McNitt-Gray, Technical Note: Design and implementation of a high‐throughput pipeline for reconstruction and quantitative analysis of CT image data. Medical Physics 46(5), 2310-2322. (2019)
- (link)(pdf) John Hoffman, Frédéric Noo, Stefano Young, Scott Hsieh, Michael McNitt-Gray, Technical Note: FreeCT_ICD: An open-source implementation of a model-based iterative reconstruction method using coordinate descent optimization for CT imaging investigations. Medical Physics 45(8), 3591-3603. (2019)
- (link)(pdf) Thomas Martin, John Hoffman, Jeffrey Alger, Michael McNitt-Gray, Danny Wang, Low‐dose CT perfusion with projection view sharing. Medical Physics 45(1), 101-113. (2017)
- (link)(pdf) John Hoffman, Stefano Young, Frédéric Noo, Michael McNitt-Gray, Technical Note: FreeCT_wFBP: A robust, efficient, open‐source implementation of weighted filtered backprojection for helical, fan‐beam CT. Medical Physics 43(3), 1411-1420. (2016)
Selected Abstracts
- (poster) Morgan A. Daly, John M. Hoffman, Using fully synthetic training data to automate clinical CT-ACR phantom analysis. Annual Meeting of the RSNA, Chicago, IL, (December 2023)
- (link) Spencer Welland, Grace H.J. Kim, Anil Yadav, John M. Hoffman, Will Hsu, Matthew S. Brown, Elham Tavakkol, Kambiz Nael, Michael F. McNitt-Gray, Assessing variability in non-contrast CT for the evaluation of stroke: the effect of CT image reconstruction conditions on AI-based CAD measurements of ASPECTS value and hypodense volume. SPIE Medical Imaging 2024, San Diego, CA, (February 2024)
- (link) Morgan A. Daly, John M. Hoffman, et al (13 additional authors), Harmonizing quantitative imaging feature values in CT using image quality metrics as a basis . SPIE Medical Imaging 2024, San Diego, CA, (February 2024)
- (link) Joshua H. Genender, John M. Hoffman, A single-file implementation of the DICOM-CT-PD raw projection data storage format. SPIE Medical Imaging 2024, San Diego, CA, (February 2024)
- (link) John Hoffman, Frédéric Noo, Michael McNitt-Gray, An analytic, physics-based approach to scoring emphysema in lung CT patients. SPIE Medical Imaging, San Diego, CA, (February 2023)
- John Hoffman, Scott Hsieh, Frédéric Noo, Michael McNitt-Gray, FreeCT ICD: Free, Open-Source MBIR Reconstruction Software for Diagnostic CT. The Fifth International Conference on Image Formation in X-Ray Computed Tomography, Salt Lake City, UT, (May 2018)
- John Hoffman, Frédéric Noo, Michael McNitt-Gray, Influence of tube current modulation on Noise Statistics of reconstructed images in low-dose lung cancer CT screening. AAPM Annual Meeting, Denver, CO, (August 2017)
- John Hoffman, Grace Hyun J. Kim, Jonathan Goldin, Matthew Brown, Michael McNitt-Gray, A Pilot Study Evaluating the Robustness of Density Mask Scoring (RA-950), a Quantitative Measure of Chronic Obstructive Pulmonary Disease, to CT Parameter Selection Using a High-Throughput, Automated, Computational Research Pipeline. AAPM Annual Meeting, Denver, CO, (August 2017)
- John Hoffman, Frédéric Noo, Stefano Young, Michael McNitt-Gray, Tailoring TCM Schemes to a Task: Evaluating the Impact of Customized TCM Profiles on Detection of Lung Nodules in Simulated CT Lung Cancer Screening. AAPM Annual Meeting, Washington, DC, (August 2016)