What it does
Allows pathologists from around the world to upload, view, and annotate images. Pathologists need to understand the virus so that vaccines and therapies can target the disease mechanisms. Once hallmarks and mechanisms of the disease are identified, AI can be used to support the evaluation of new cases to diagnose or direct patient management.
How we built it
Several teams and companies have built the platforms. What is needed now are templates for collecting meta data and tissue evaluations, standard storage and data formatting methods, and a software package for querying the repository.
Challenges we ran into
COVID19 Pathophysiology must be evaluated. To study this, researchers and physicians will need a large repository of study samples.
Collecting samples on autopsy is challenging:
- Victims are being buried without any autopsy or specimen harvesting because space is needed in hospitals and there is risk that the deceased can transmit the virus.
- Some countries are not allowing autopsy due to rationing of PPE.
Creating the repository infrastructure is challenging:
- We are communicating with our international network to find sites that are doing autopsies. We have managed to obtain some samples to use to start building the specimen intake pipeline. There is currently no common messaging, and getting the right people’s attention is hard- everyone is very busy right now.
- Industry partners need some kind of agreement for sharing images ensuring clarity and fairness. Partners are willing to do this for the public good but want to make sure everyone plays by the same rules.
- Establishing the ethical materials (IRBs) is also a challenge. We expect that IRBs will allow latitude and attempt to expedite decisions.
Building a cohesive approach is challenging:
- In pathology, tissue sample collection varies greatly by human factors. We will need to build a standardized approach to collection locations, methods, tissue staining, etc.
- Data storage is often different for each slide scanner or metadata collected. We need to ensure that all data may be collected and stored in a common format
- We need to make sure that the data will be query-able by all. The solution should be open source, to allow for all parties to access the data, given the correct PII precautions and other data sharing requirements are met.
Accomplishments that we are proud of
Being a part of the alliance of a diverse set of stakeholders that include academic subject matter experts (clinical and technology), industry, US government (NIH, FDA), international consortia, and experts from the World Health Organization.
What we learned
People want to help.
What's next for COVID19 Digital Pathology Database
Get digital pathology images and create the templates for meta data and pathology evaluations. Create the research administrative materials. Spread the word.
You can see the platforms for viewing and evaluating digital pathology images by checking out two data-collection portals from a current project:
- https://wolf.cci.emory.edu//camic_htt/apps/landing/landing.html (need google account)
More information about the project:
- Data-collection event info: https://ncihub.org/groups/eedapstudies/wiki/HighThroughputTruthingYear3
- Data-collection webinar video: https://ncihub.org/groups/eedapstudies/wiki/HighThroughputTruthingYear3/File:20200219_HTTdataCollectionWebinar_Trim.mp4
pathpresenter, the-cancer-genome-atlas, the-cancer-imaging-archive