The Medical Imaging De-Identification workshop is being offered to examine the state of the science for the de-identification of publicly released medical images. The primary emphasis is on medical images with accompanying data elements, especially those encoded in formats in which the data elements are embedded, particularly DICOM. The sessions described below make up the workshop.
Please register here.
DAY 1, May 22, 2023 (10:00 am - 2:00 pm EDT)
Welcome and Opening Comments
10:00 am - 10:10 am EDT
Keyvan Farahani, PhD, National Heart, Lung, and Blood Institute & National Cancer Institute, National Institutes of Health
Session 1: Report of the MIDI Task Group - Best Practices and Recommendations
10:10 am - 11:00 am EDT
In this session, David Clunie, chair of the MIDI Task Group, will summarize the best practices and recommendations included in the task group's report, recently available in pre-print, followed by a question and answer period.
Session Chair: David Clunie, MBBS, PixelMed
10:10 am - 10:50 am David Clunie, MBBS, PixelMed
Summary of the MIDI Task Group Report
10:50 am - 11:00 am Discussion
Session 2: Tools for Conventional Approaches to De-Identification
11:00 am - 11:50 pm EDT
In this session, speakers will share methods currently in use for medical de-identification.
Session Chair: Fred Prior, PhD, University of Arkansas for Medical Sciences
11:00 am - 11:10 am Fred Prior, PhD, University of Arkansas for Medical Sciences
Setting the Stage
11:10 am - 11:20 am Michael Rutherford, MS, University of Arkansas for Medical Sciences
The Tools of TCIA: Standardizing Zero-Tolerance De-identification
11:20 am - 11:30 am Stephen Moore, MS, Washington University School of Medicine in St. Louis
XNAT Platform: Image De-identification
11:30 am - 11:50 am Panel Discussion
Break 11:50 am - 12:00 pm
Session 3: International Approaches to De-Identification
12:00 pm - 12:40 pm EDT
This session will focus on requirements for de-identification outside of the United States.
Session chair: William Parker, MD, University of British Columbia
12:00 pm - 12:10 pm William Parker, MD, University of British Columbia
12:10 pm - 12:20 pm Haridimos Kondylakis, PhD, Institute of Computer Science, Foundation of Research & Technology (FORTH)
Data Infrastructures for AI in Medical Imaging: A report on the experiences of five EU projects
12:20 pm - 12:30 pm Christian Ludwigs, MSc, Aigora GmbH
Legal Framework and Best Practices for Medical Image De-Identification in the EU
12:30 pm - 12:40 pm Discussion
Session 4: Industry Panel on Image De-Identification
12:40 pm - 1:50 pm EDT
This session, largely a panel discussion, will feature flash presentations by industry groups and discuss their innovative approaches to image de-identification.
Session chair: Juergen Klenk, PhD, Deloitte Consulting
Panelists:
- Abraham Gutman, MS, AG Mednet
Advances in Medical Imaging De-Identification and the Impact of Regulatory Constraints - Dan Marcus, PhD, Flywheel
The Flywheel Platform for Intelligent Image Anonymization - Bob Lou, MD, Google
Medical Imaging De-identification on Both Images and Text using AI Models - Lawrence (Tony) O’Sullivan, MS, IBIS
Optimizing and Automating Radiology - Jiri Dobes, PhD, John Snow Labs
Day 1 Closing Remarks
1:50 pm - 2:00 pm EDT
David Clunie, MBBS, PixelMed
DAY 2, May 23, 2023 (10:00 am - 2:00 pm EDT)
Welcome and Recap
10:00 am - 10:10 am EDT
David Clunie, MBBS, PixelMed
Session 5: Pathology Whole Slide Image De-Identification
10:10 am - 11:00 am EDT
In this session, researchers will discuss pathology whole slide image de-identification.
Session chair: Adam Taylor, PhD, Sage Bionetworks
10:10 am - 10:20 am Adam Taylor, PhD, Sage Bionetworks
10:20 am - 10:30 am Tom Bisson, PhD, Charité Universitätsmedizin Berlin
Anonymization of Whole Slide Images in in Histopathology for Research and Education
10:30 am - 10:40 am David Gutman, MD, PhD, Emory University
Image DePHI and the DSA: Open Source tools for Histology Image De-Identification
10:40 am - 11:00 am Panel Discussion
Session 6: De-facing
11:00 am - 12:00 pm EDT
This session will focus on balancing the risks of removing potentially reconstructable facial information in head and neck cross-sectional images, also called de-facing, with the diminished utility of these images caused by restricted access to them.
Session chair: Ying Xiao, PhD, Hospital of the University of Pennsylvania
11:00 am - 11:10 am Ying Xiao, PhD, University of Pennsylvania
11:10 am - 11:20 am Christopher Schwarz, PhD, Mayo Clinic
Face Recognition and De-Identification of Research Brain Images with mri_reface
11:20 am - 11:30 am Douglas Greve, PhD, MGH/Harvard
MIDEFACE: Minimally Invasive Defacing
11:30 am - 11:50 am Panel Discussion
Break 11:50 am - 12:00 pm
Session 7: The Role of AI in Image De-Identification
12:00 pm - 12:50 pm
In this session, researchers discuss the utility of AI algorithms in image de-identification.
Session chair: Judy Wawira Gichoya, MD, Emory University
12:00 pm - 12:10 pm Judy Wawira Gichoya, MD, Emory University
12:10 pm - 12:20 pm George Shih, MD, Weill Cornell Medical College
Pixel De-Identification Using AI
12:20 pm - 12:30 pm Adrienne Kline, MD, PhD, Northwestern University
PyLogik: An open-source resource for medical image de-identification
12:30 pm - 12:50 pm Panel Discussion
Session 8: NCI MIDI Datasets and Pipeline
12:50 pm - 1:50 pm
This session will present CBIIT/NCI Medical Image De-Identification Datasets and Pipeline
Session chair: Keyvan Farahani, PhD, National Heart, Lung, and Blood Institute & National Cancer Institute, National Institutes of Health
12:50 pm - 1:00 pm Keyvan Farahani, PhD
1:00 pm - 1:10 pm Fred Prior, PhD, University of Arkansas for Medical Sciences
Synthetic Data for De-Identification Testing
The MIDI Datasets
1:10 pm - 1:20 pm Ben Kopchick, PhD, Deloitte Consulting
Building a cloud-based MIDI pipeline
1:20 pm - 1:50 pm Panel Discussion
Closing Remarks
1:50 pm - 2:00 pm EDT
David Clunie, MBBS, PixelMed
Keyvan Farahani, PhD, NHLBI & NCI, NIH
