Subject of this announcement: I took a snapshot of the attendees near the beginning of the meeting. The list is at the bottom.
All teams introduced themselves and their objectives. Most of the introductions are reflected in the team objectives. A few groups have taken ownership of these pages by updating them with more details. Everyone should do the same, please update your pages appropriately.
Here is Brandon’s grouping of objectives discussed.
- Collect reference data for machine learning.
- Dig a little deeper to morphometric information.
- Scale up collection of reference data. Need more. Crowd-sourcing?
- Understand reader variability and use it to characterize quantitative labels (again for machine learning). Instead of binary decisions (mitotic figure or not), allow the label of a cell to scale from 0 to 100.
- Benchmark equipment, acceptance criteria, QA/QC
Here are some other issues that came up.
- The IIT group has some big objectives, check them out. They are trying to find a solution to their shortage of pathologists. I think Mohan said it was about 100 million patients for every 10 pathologists. They hope image analysis will play a big role in serving the large population, determining slide-prep quality, and evaluating criteria for slide-prep quality.
- Mitotic classification, search and enumeration, seem to be tasks that provide a good starting point. Other tasks to follow.
- Brandon will visit MSKCC in June to loan and install an eeDAP system for data collection. CSHL may join the visit to learn. Others are invited.
- Some objectives may not require evaluation in microscope mode, but collecting data with the microscope mode could supply a useful baseline for performance.
- The point of this group is to share: evaluations, images, protocols, expertise.
- One of the first agreements that we should make is that we honor the “right to publish first” of the investigator that collects data.
- We should also create an agreement that each of us is expected to announce plans to publish anything related to our discussions and that the group should be given time to review publication plans, provide feedback, and suggest coauthors as appropriate.
Pathology Informatics Plan
- I will notify WSI WG. I also hope to get them involved providing feedback on eeDAP study plans.
- We will have a meeting on Sunday before the main part of Pathology Informatics Summit.
- 2 presenters. I really hope I can convince two of you to present.
- 2-4 posters. This is not that hard. Please plan on creating a poster updating the WSI WG with your eeDAP study plans.
- Start detailing eeDAP study plans and protocols. Let’s see if we can’t start with the STARD checklist and tune it to our needs
- In the near future, I would like to partner teams for protocol development, this will include teams in the eeDAP studies group and members of the larger WSI WG.
- GitHub site for eeDAP: https://github.com/DIDSR/eeDAP
- Download eeDAP from the release page: https://github.com/DIDSR/eeDAP/releases
- Documentation at: http://didsr.github.io/eeDAP/
- Please start on your own using the documentation. Let us know if you need more help. Let us know the documentation deficiencies and issues at https://github.com/DIDSR/eeDAP/issues
- Qi Gong and I are ready to help. We can get on the phone to help, share screens.
Technical Performance of eeDAP
We didn’t get to this during the T-con, but one of the things that needs to be done is to characterize registration precision.
I am sure I missed some folks. Please add accordingly.
- FDA team: Brandon Gallas, Qi Gong, Weijie Chen, Marios Gavrielides
- NCI team: Stephen Hewitt, Mark Simpson
- IIT Madras team: “Mohan” Mohanasankar Sivaprakasam, Jaikishan Jayakumar
- MSKCC team: Joe Sirintrapun, Yukako Yagi
- Philips team: Liselotte Kornmann, Prarthana Shrestha, Mischa Nelis
- Inspirata team: Mark Lloyd, James Monaco
- CSHL team: Partha Mitra