Skip to content

Deploy Working Group

David Bericat edited this page Feb 8, 2022 · 21 revisions

MONAI Deploy Working Group

MONAI Deploy aims to become the de-facto standard for developing, packaging, testing, deploying and running medical AI applications in clinical production.

Expanding MONAI to cover the "end-to-end AI lifecycle"

Project MONAI was created with the goal of “accelerating the pace of research and development by providing a common software foundation and a vibrant community for medical imaging deep learning”.

There is a lot of research in the development and training of AI models by data science teams across the world, and the discipline is becoming mature thanks to initiatives like MONAI.

But in order to realize the value of all that effort and investments, the medical community has to naturally evolve to put all that new innovation seamlessly integrated within the different clinical workflows and systems, supporting and augmenting physicians, and freeing up their time to spend more time providing care to patients. That jump today is simply too high, almost a leap of blind faith. When we are talking about people’s lives, disease detection, diagnosis and treatment, not knowing how the results will affect care is simply unacceptable.

Therefore, we see an opportunity to create a set of intermediate steps, where researchers and physicians can build confidence in the techniques and approaches used with AI, deploy and connect with their research medical devices and systems in a controlled environment (i.e. Research PACS), and consume the results they need at the speed they are ready to, iterating over time until the overall AI inference infrastructure is ready to move to clinical environments, with a high certainty the transition will go smoothly.

We would love project MONAI to be the community where this idea develops, expanding its scope from AI from model training to QA evaluation first, and then to clinical deployments.

MONAI Deploy WG - scope diagram

Focus

MONAI Deploy builds on the foundation set by MONAI.

Where MONAI is focused on training and creating models, MONAI Deploy is focused on defining the journey from research innovation to clinical production environments in hospitals. Our guiding principles are:

  • Implementation mindset. Create tangible assets: tools, applications and demos/prototypes.
  • Radiology first, then other modalities like Pathology.
  • Interoperability with clinical systems. Starting with DICOM, then FHIR.
  • Central repository to facilitate collaboration among institutions.

Status

MONAI Deploy was released at MICCAI 2021 and was part of the MONAI 2021 Bootcamp. Since then we have released several versions of some of the sub-systems, while others are being actively developed. Please check out the next section.

Sub-systems

First versions include:

Future versions will include:

Community

To participate, please join the MONAI Deploy WG weekly meetings on the calendar. All the recordings and meeting notes since day zero can be found at MONAI Deploy WG master doc

Join our Slack channel or join the conversation on Twitter @ProjectMONAI.

Ask and answer questions over on MONAI Deploy's GitHub Discussions tab or MONAI's GitHub Discussions tab.

Members

Group Leads

  1. Haris Shuaib (Guy’s & St Thomas NHS Foundation Trust)
  2. David Bericat (NVIDIA)

Working Group institutions

KCL, GSTT, AI Centre, Answer Digital, Mayo Jacksonville, Vanderbilt, MGB, UCSF, UCSD, DKFZ, Stanford, NIH, Kitware, IBM, Quantiphi, VA, Philips, Nuance, IHU, MeThinks, NVIDIA

Meeting Notes

MONAI Deploy WG - master doc

Clone this wiki locally