MIDRC newsletter.

Newsletter.

Stay in touch through the MIDRC newsletter! Sign up to receive the newsletter.

Issue 10, December 2023:

*|MC:SUBJECT|*
Welcome to the December 2023 issue of the
Medical Imaging and Data Resource Center (MIDRC) newsletter!
The ARPA-H BDF is an initiative to develop technologies for an easily deployable, multi-modal, multi-scale, connected data ecosystem for biomedical data. ARPA-H is enlisting multiple performers who are experts in their fields to build components of the ARPA-H BDF Toolbox and MIDRC was selected to provide domain expertise and data commons technology development in medical imaging.

MIDRC's expertise stems from developing a data commons in which contributed imaging data are harmonized and vetted by clinical, artificial intelligence (AI)/machine learning (ML), and medical imaging & data science domain experts from the collaborative communities of AAPM, ACR, RSNA, and Gen3. The public open data commons was created “at scale” and is interoperable with other data commons to enable multi-modal, multi-omics research as well as rigorous statistical evaluations.

MIDRC’s involvement in the ARPA-H BDF Toolbox will include development and deployment of medical imaging data commons architectures and resources to support various diseases and applications for the toolbox. Stay tuned on what's to come! 
RSNA's 109th Scientific Assembly and Annual Meeting, was held in-person in Chicago, IL from November 26-30, 2023. MIDRC presented a number of insightful presentations, engaging workshops, and exhibit. 
To name a few: 
 
-MIDRC: Data for Your AI
-MIDRC: Using a Massive Multi-Institutional Open Imaging Data Commons for Machine Learning Research
-MIDRC: Building and using AI-ready datasets from a massive open data commons
-The MIDRC Experience with Curating DICOM Images for use in AI Research
-Assessment of 3D Image Quality Variations in the MIDRC Chest CT Database: A Multi-institutional Study of Imaging Systems and Protocols
 
The well-trafficked MIDRC exhibit in the Lakeside Learning Center highlighted the accomplishments of our work since the creation of MIDRC back in August of 2020. Check out a few pictures from the meeting below - if you were not able to attend this year, we hope to see you next year!

As the year comes to a close, we wanted to celebrate the remarkable achievements of the MIDRC collaborative groups by highlighting their accomplishments!

Our year of achievements:
  • Increased ingestion to more than 300,000 imaging studies with over 159,000 imaging studies available on the public user portal (with the remaining currently undergoing curation).
  • Trained dozens of new helper-AI models to classify image attributes from pixel-level data, i.e., images
  • Established a process and infrastructure to allow collaborators an easy mechanism for sharing complete or partial data sets with MIDRC.
  • Provided and implemented an approach for mapping the DICOM header information of MIDRC images to a structured list that can be used in the MIDRC data model for searching and building image cohorts by users. The mapping is based on the LOINC/RSNA Radiology Playbook, which is a proposed unified standard for radiology procedure names.
  • Created a decision tree tool (MetricTree), available to the public on our website (midrc.org) to recommend appropriate performance metrics for different machine learning tasks.
  • Created a bias awareness tool, available to the public at midrc.org, and MIDRC's focus on bias and diversity resulted in two recent MIDRC publications that can be found here, and here
  • Helmed our 3rd Grand Challenge, the MIDRC mRALE Mastermind Challenge. Finalists received cash awards, as well as the opportunity to have their resultant code supported by MIDRC through the FDA regulatory process. 

Explore MIDRC's YouTube channel for indispensable resources and expert insights. We currently have 40 videos with more on the way!
    These seminars are an opportunity for members of the medical community at-large to hear directly from our team. Our virtual seminars are free and open to everyone, click here for the 2024 registration link!

Many algorithms are in development within MIDRC. Please see our links (both at midrc.org, the GitHub direct link, and the Huggy Face) for a list of algorithms/models/code that are publicly available right now, many more to come!

Multiple types of data annotations are now available in the MIDRC data portal, searchable in the data explorer as well as through custom Jupyter notebooks.
Sign Up Here
Website
YouTube
Twitter
LinkedIn
If you are interested in becoming a MIDRC partner please contact kpizer@bsd.uchicago.edu