September 13, 2022 (Dev Meeting)

Agenda:

#	Owner	Title	Time (min)
1	Behrooz	Final discussion for v1.0	10
2	Behrooz	Lizard dataset licensing issue	10
3	Jonny	Integration issues with MONAI Label for QuPath and DSA	20

Attendees

David Manthey, Behrooz Hashemian, Shan Reza, Jonny Hancox, Lee Cooper

Notes

• Behrooz explained the final changes for MONAI v1.0 and implication of any future change that needs to be backward compatible.
• We discussed about the change of location argument/attribute order to be compatible with OpenSlide order (x, y) instead of (y, x). We evaluated if it is worth to change the current convention in MONAI WISReader. The consensus was to keep it as is since there is no time to fully test it before the release and if we want to do it after the release with an order flag (which needs to be backward compatible), it will create more confusion for user as they need to check if this flag is set or not.
• Shan mentioned that the Lizard dataset and many other dataset in PathLake will change and will have a more permissive terms of use.
• Jonny raised the fact that there is no practical way of using the Pathology POC for MONAI Label in either QuPath or DSA. Specifically, in DSA, it would be useful to be able to use the NuClick algo to correct inaccurate cell-level inference, but the current implementation of DSA only provides a means of executing a plug in such as Nuclick via a Docker container instantiation. This is impractical for Nuclick because each cell click would require the creation of a new container and then its destruction, which would involve too much latency. Instead there needs to be a way of accomplishing this without spawning a container for each click. David indicated that this was on the roadmap and he expects to get round to addressing it in the autumn.

Action items:

PIC	Item
Behrooz	finalize the pathology features for v1.0 release
Behrooz	close the related issues with location order
Jonny	Discuss about the missing pieces of MONAI Label integration with the stakeholders