Project MONAI Tutorial

MICCAI 2023

October 8th @ 8:00 AM to 3:00 PM (PDT time)

Developing for the Medical AI Project Lifecycle with MONAI

Background

MONAI is the Medical Open Network for AI. 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.

MONAI is targeted for both the medical image computing (MIC) and the computer-assisted intervention (CAI) communities. It is an extension to PyTorch that includes specialized deep learning methods for diagnostic radiology, surgical planning, medical robotics, image-guided interventions, and nearly every other aspect of patient care.

MONAI provides a customizable interface to bring components to existing training pipelines or perform end-end training. Its goal is to increase efficiency and collaborations amongst researchers by providing domain-optimized foundational capabilities and standardization around algorithms’ benchmarking, evaluation, and reproducibility. These goals help improve reproducibility by not only being used as a framework for development but comparison, evaluation, and benchmarking by bringing together the community and organically converging on a standard set of best practices.

Learning Goals

After participating in this session, attendees should be able to:

MONAI Core:

  • Integrate the main components from MONAI Core in your own workflow, including Transforms, Metrics, Losses, Network Architectures, and our new MetaTensor data structure
  • Use MONAI Bundles to help simplify your model development lifecycle throughout MONAI
  • Utilize the MONAI Model Zoo and understand how to contribute a model
  • Understand the new MONAI Federated Learning API
  • Find the latest state-of-the-art techniques, networks, and frameworks for MONAI Core

MONAI Label:

  • Explain the benefits and setup MONAI Label
  • Extend MONAI Label to their own custom application
  • Explain the MONAI Continous Training Loop and Active Learning options

MONAI Deploy:

  • Discuss how MONAI Deploy is bridging the gap from research innovation to clinical production
  • Implement building, verifying, and packaging a deploy-ready AI Application (MAP) using MONAI Deploy App SDK
  • Summarize how MONAI Deploy and the AIDE Platform are being used today

Speakers

Michael Zephyr

Michael Zephyr

Technical Product Manager, NVIDIA

Seyed-Ahmad Ahmadi

Seyed-Ahmad Ahmadi

Senior Solution Architect, NVIDIA

Andres Diaz-Pinto

Andres Diaz-Pinto

Technical Marketing Engineer, NVIDIA

Vikash Gupta

Vikash Gupta

Assistant Professor of Radiology, Mayo Clinic

Annika Reinke

Annika Reinke

Post-Doctoral Researcher, DKFZ

Mark Graham

Mark Graham

Research Fellow, King's College London

Andriy Myronenko

Andriy Myronenko

Senior Researcher, NVIDIA

Nobuhiko Hata

Nobuhiko Hata

Professor, Harvard Medical School and Director of Innovation, Radiology Research Administration, Brigham and Women's Hospital

Franklin King

Franklin King

Senior Research Engineer, Brigham and Women's Hospital

Marc Edgar

Marc Edgar

Senior Alliance Manager for Medical Instruments, NVIDIA

Agenda

8 am - 10 am

MONAI Intro

Welcome and Introduction to MONAI

MONAI Core

  • Intro to MONAI Core
  • End-to-End Training with MONAI
10 am - 10:30 am

Coffee Break

Enjoy some refreshments

10:30 am - 12:30 pm

MONAI Label

  • Intro to MONAI Label
  • Demo: 3D Slicer/OHIFv3 + MONAI Label

MONAI Deploy

  • Overview of MONAI Deploy
12:30 pm - 1:30 pm

Lunch

Enjoy your lunch break

1:30 pm - 3:30 pm

MONAI Research topics:

  • Metrics Reloaded
  • Generative AI
  • Auto3DSeg
3:30 pm - 4:00 pm

Coffee Break

Enjoy some refreshments

4 pm - 5:10 pm

MONAI Stream

Introduction to MONAI Stream

Contributing to MONAI

Learn how to contribute and closing remarks

Closing Remarks

Want to learn more about MONAI today?

Below you'll find links to the MONAI Website, Tutorials Repo, GitHub Organization, and Slack Channel where you can start to get involved and contribute today.