MONAI Leaps Forward with AutoML-Powered Model Development and Cloud-Native Deployments

Project MONAI continues to expand its end-to-end workflow with new releases and a new component called MONAI Deploy Inference Service.

NVIDIA logo
NVIDIA
•
Michael Zephyr
•3 min read•intermediate•
--
•View Original
Active LearningAutoMLHelmKubernetesREST APISupervised Learning

Overview

Project MONAI has made significant advancements with the release of MONAI v0.8, MONAI Label v0.3, and MONAI Deploy App SDK v0.2, along with the introduction of the MONAI Deploy Inference Service. These updates enhance model development and deployment capabilities, particularly in cloud-native environments using Kubernetes.

What You'll Learn

1

How to utilize Self-Supervised Learning techniques in MONAI Core v0.8

2

Why Multi-Instance Learning is important for medical imaging tasks

3

How to deploy MONAI Application Packages using the MONAI Deploy Inference Service

Prerequisites & Requirements

  • Understanding of Kubernetes and cloud-native deployments
  • Familiarity with medical imaging concepts(optional)

Key Questions Answered

What are the new features in MONAI Core v0.8?
MONAI Core v0.8 introduces Self-Supervised Learning and Multi-Instance Learning support, along with a differential search framework called DiNTS for accelerating Neural Architecture Search (NAS) in large-scale 3D image sets. These features enhance the learning capabilities of the framework.
How does MONAI Deploy Inference Service enhance model deployment?
The MONAI Deploy Inference Service allows for the deployment of MONAI Application Packages (MAPs) as cloud-native microservices within a Kubernetes cluster. It facilitates resource provisioning, input handling via REST API, and output delivery to clients, streamlining the deployment process.
What improvements does MONAI Label v0.3 offer?
MONAI Label v0.3 enhances multilabel segmentation support using DynUNet and UNETR networks, introduces multi-GPU training for better scalability, and includes user experience improvements for active learning, making it easier for developers to implement these features.

Technologies & Tools

Some links below are affiliate links. We may earn a commission if you make a purchase.

Orchestration
Kubernetes
Used for deploying MONAI Application Packages as cloud-native microservices.
Framework
Monai
Framework for medical imaging AI model development.

Key Actionable Insights

1
Leverage the new Self-Supervised Learning features in MONAI Core v0.8 to improve model performance on unlabeled datasets.
This approach is particularly useful in medical imaging where labeled data is scarce, allowing models to learn from vast amounts of unlabeled data.
2
Utilize the MONAI Deploy Inference Service to transition your models from local environments to production with ease.
By deploying models as microservices in Kubernetes, you can ensure scalability and manageability, which is crucial for applications requiring high availability.

Common Pitfalls

1
Failing to properly configure Kubernetes resources for MONAI Deploy can lead to deployment failures.
Ensure that you understand the resource requirements of your MONAI Application Packages and configure your Kubernetes cluster accordingly to avoid issues during deployment.

Related Concepts

Cloud-native Deployments
Neural Architecture Search
Self-supervised Learning
Multi-instance Learning