Click here to

Session: Medical Imaging and Data Resource Center: Imaging in Covid [Return to Session]

Medical Imaging and Data Resource Center: Imaging in Covid

M Giger1*, P Kinahan2*, M McNitt-Gray3*, K Myers4*, S Armato5*, (1) University of Chicago, Chicago, IL, (2) University of Washington, Seattle, WA, (3) David Geffen School of Medicine at UCLA, Los Angeles, CA, (4) Office of Science and Engineering Laboratories, FDA, Silver Spring, MD, (5) The University of Chicago, Chicago, IL


1:00 PM Overview of MIDRC and COVID-19 AI Research Projects - M Giger, Presenting Author
1:20 PM Image Quality and Harmonization in MIDRC - P Kinahan, Presenting Author
1:40 PM Rigorous Validation, Appropriate Metrics, and Task-based Distribution of Data for Evaluating COVID-19 AI - M McNitt-Gray, Presenting Author
2:00 PM Rigorous Validation, Appropriate Metrics, and Task-based Distribution of Data for Evaluating COVID-19 AI - K Myers, Presenting Author
2:20 PM Grand Challenges through MIDRC - S Armato, Presenting Author
2:40 PM Q&A & Panel Discussion - M Giger, Presenting Author

SU-CD-TRACK 3-0 (Sunday, 7/25/2021) 1:00 PM - 3:00 PM [Eastern Time (GMT-4)]

The COVID-19 pandemic presents a critical public health crisis. Essential biomedical research and development is needed to urgently address surveillance and early detection of COVID-19 resurgence; differential diagnosis of COVID-19 patients; prognosis; and prediction and monitoring of response for use in patient management. Since the start of the pandemic, medical imaging has played key roles in these patient monitoring and assessment tasks. In response to this need, leaders from the RSNA, ACR, and AAPM along with NIBIB have jointly developed the Medical Imaging and Data Resource Center (MIDRC) for rapid, flexible, and large collection of well-curated imaging studies to facilitate development of and externally validate machine intelligence (AI) algorithms for the various tasks described above. MIDRC is a multi-group, two-year NIBIB-funded project hosted at the University of Chicago and coordinated among 23 institutions including two intake portals operated by RSNA-RICORD and ACR-CIRR that integrate quality control and data harmonization. A public access, exploration, and analysis portal and commons is being developed on the Gen3 data ecosystem. Twelve collaborative research projects exist to expedite translation of machine learning and AI from scientific findings and technical resources to public dissemination and clinical benefit. In this symposium, we will review the status of MIDRC, the role of AAPM’s Data Science Committee, and the important role that medical physicists have in MIDRC for quality assessment and data harmonization, machine intelligence assessment through metrology and task-based evaluations, and the conduct of grand challenges.

Learning Objectives:
1. Learning objective 1: Learn about MIDRC and the AAPM's leadership and development roles in it.
2. Learning objective 2: Understand the need for large, well-curated medical imaging datasets to facilitate development of AI and to externally validate such algorithms in order to enable translation to clinical use.
3. Learning objective 3: Understand the essential medical physics research needed to address the role of medical imaging for monitoring, diagnosis, and assessment of the COVID-19 patient.

Funding Support, Disclosures, and Conflict of Interest: Funding for MIDRC is from NIBIB COVID-19 Contract 75N92020D00021



Computer Vision, Quality Assurance, Chest Radiography


IM- X-Ray: General (Most aspects)

Contact Email