FAQs

General topics

+ Are images available for download at the moment?

Yes! Go to data.mirc.org to access our data. Data collection is ongoing so check back often for updates.

+ What is the Medical Imaging and Data Resource Center (MIDRC)?

MIDRC is a multi-institutional initiative driven by the medical imaging community and aimed at accelerating the transfer of knowledge and innovation in the current COVID-19 pandemic. MIDRC, funded by the National Institute of Biomedical Imaging and Bioengineering (NIBIB) and hosted at the University of Chicago, is co-led by the American College of Radiology® (ACR®), the Radiological Society of North America (RSNA), and the American Association of Physicists in Medicine (AAPM). The aim of MIDRC is to foster machine learning innovation through data sharing for rapid and flexible collection, analysis, and dissemination of imaging and associated clinical data by providing researchers with unparalleled resources in the fight against COVID-19. MIDRC is creating a large-scale de-identified dataset of medical images of COVID-19 patients, and includes support for research projects that capitalize on the valuable dataset it creates. This dataset will be made available, through a Gen3 portal, to the public to support the development of AI algorithms that may lead to better characterization of COVID disease and guide future clinical interventions.

+ Why was the MIDRC founded?

The MIDRC was founded to collect and provide large numbers of curated medical images with known characteristics to enable AI algorithm development. In a sense like the approach used by massive publicly-available (non-medical) image data sets like ImageNet, which have driven progress in AI outside of medicine. Machine learning for medical image analysis is often hampered by limited dataset sizes, inconsistent image quality, and a lack of solid ground truth. With the recent emergence of the novel coronavirus, the development of AI algorithms for COVID-19 detection, diagnosis, prognosis, and predicting treatment response has been challenged by a lack of sufficient data for training and testing algorithms, has often been based on patients unrepresentative of the population, or has used images of inferior or unknown image quality.

+ Who founded the MIDRC?

The MIDRC was founded in August 2020 by a joint effort of the American Association of Physicists in Medicine (AAPM), the American College of Radiology (ACR), and the Radiological Society of North America (RSNA). In early 2020 the COVID-19 pandemic emerged in the US, which provided an urgent need and direction (or use case) for the MIDRC. MIDRC involves the AAPM, ACR, RSNA, NIBIB, and the University of Chicago (as the PI hosting site) and over 20 research universities and professional societies.

+ There are other COVID-19 image data sets being created – what makes MIDRC different?

MIDRC has a unique ability to scale its data collection activities because of the close collaboration of three major radiology/medical physics professional societies, as well as provide COVID-19 data including images, associated clinical/laboratory data, and associated annotations of disease involvement. MIDRC also possesses a unique level of expertise related to medical imaging and is therefore pursuing an approach to optimize the value of combining medical images with the clinical data typically being analyzed by other COVID-19 groups. Most other efforts are focusing on the broad range of all patient clinical, laboratory and diagnostic data, and contact tracing. The MIDRC possesses a unique level of expertise related to medical imaging and is therefore pursuing an approach which will look to optimize the value of combining medical images with the clinical data typically being analyzed by other COVID-19 groups. Specifically, MIDRC will take advantage of the DICOM format used in medical imaging, which is rich in content, rigorously defined, and offers additional data which can be analyzed in conjunction with the images and the clinical data.

Data types

+ What image format is used?

The MIDRC uses the nearly universally adopted Digital Imaging and Communications in Medicine (DICOM) standard for medical images. It is the international standard for transmitting and storing medical images and related information and uses a fundamentally different approach compared to simpler images, such as JPEG, TIFF, or GIF. DICOM is used clinically and by all imaging hardware and software manufacturers, and provides rich information on acquisition and imaging protocol data as well as the images. While there are simpler image formats used in some forms of research, the MIDRC images are retained in DICOM format to preserve information. Many software tools are available for interpreting and converting DICOM images to other formats. A good place to look for open source software is the I Do Imaging website. More details on DICOM and other medical image formats can be found at David Clunie's Medical Image Format Site. Note that some legacy datasets are in the NIfTI format rather than DICOM.

+ What about data other than medical images?

Compared to DICOM data, other forms of medical information, e.g. Electronic Health Records (EHRs), vary considerably. Each has often been collected for a different purpose, resulting in sometimes wide variances of organization, format, and terminology from source to source. These include demographics, symptoms, lab test results, procedures, medications, medical conditions, physical measurements and more. The Observational Medical Outcomes Partnership (OMOP) model is being adopted by some groups such as the NIH-support National COVID Cohort Collaborative (N3C) to allow users to generate evidence from a wide variety of sources. MIDRC expects to develop collaborations with such federally funded efforts (such as N3C, All of Us, BioData Catalyst, etc). In addition the ACR has established a COVID Imaging Research Registry (CIRR) that will collect extensive clinical data along with medical images. The RSNA International COVID-19 Open Radiology Database (RICORD) initiative can also collect non-imaging data. Contributing sites will provide demographic information, clinical data on signs and symptoms, imaging exams, laboratory test data and outcomes. The data dictionary being used by MIDRC has been mapped against data dictionaries being used in other COVID-19 databases to harmonize across datasets.

+ What kinds of images are there?

The primary imaging collection will be chest X-ray and chest CT images at the times of diagnosis and treatment. We are also interested in imaging of the heart and brain for patients diagnosed as positive for COVID-19 given evidence that suggests involvement of other organ systems in some patients. We also intend to populate post-COVID-19 longitudinal image sets to assess the longer-term effects of COVID-19 on involved organs.

+ Is there access to treatment and outcomes information?

MIDRC is actively seeking methods to collect or link to treatment and outcomes information, as well as longitudinal tracking of patient images (including post COVID-19 infection) as they become available.

Data access and contributions

+ How do I get the data and is there a fee?

There is no fee for non-commercial use. Information on available data is listed on the MIDRC data page.

+ What if I have a technical question about the data or annotations available in the MIDRC data portal?

Please contact MIDRC data portal tech support midrc-support@datacommons.io with technical questions.

+ Are there any data use restrictions?

There are no restrictions for non-commercial use. The approaches are up to the investigators using the data. Users are required, however, to register and attest they will abide by the data use agreement. The use of MIDRC images and data in any research must be properly acknowledged in publications, presentations, or other dissemination methods. Data will not be available for download for commercial use, but for inquiries about commercial use of MIDRC, please contact us at our contact page.

+ Can I contribute images and other data from COVID-19 patients?

Yes! MIDRC is actively soliciting images and data contributions from medical centers, community hospitals, and others. Please visit the MIDRC data intake page for further information and also note that MIDRC is happy to help you throughout the process.

+ What about privacy, how are the image data de-identified?

For upload of retrospective or bulk data, contributors will deidentify all data on premise using approved software tools prior to submission to the archive. Submitted deidentified data will be embargoed temporarily for processing and curation to ensure complete eradication of protected health information (PHI). Other submission mechanisms will be supported that allow for contemporaneous transmission of data through secure channels that could contain retained PHI. Retained PHI on data submitted through the secondary mechanism will remain embargoed.

+ Are ALL of the data available?

No. The goal is to make as much data as possible publicly available. However, some data will be sequestered for validation testing.

+ Can the data be accessed from outside the United States?

Yes.

Other

+ How can I participate in the MIDRC?

Once you have registered as a MIDRC user, you will receive our news letter and other updates and can capitalize on new opportunities as they arise. This would include addition of supplemental datasets and annotations as they become available.

+ Is there funding to support my research with this data?

We encourage researchers to independently seek funding to work with MIDRC data, and would be happy to support your application with a letter of support for access to the data. If you register as a MIDRC user, you will hear about funding opportunities as they become available. There is no funding available directly from the MIDRC.

+ Who do I contact if I have questions not answered above?

Please visit our contact page if you still have questions. For tech support, email midrc-support@datacommons.io.

Last updated March 12, 2021