UNSW AI in Medical Imaging – MONAI

Medical Open Network for Artificial Intelligence (MONAI) is a PyTorch-based framework designed for deep learning in healthcare imaging. It leverages AI to deliver state-of-the-art, end-to-end workflows for healthcare imaging—from model development and research to clinical production. 

At the University of New South Wales (UNSW), several research groups focus on medical imaging. Dr Abdullah Shaikh, representing Intersect’s Advanced Analytics and AI platform at UNSW, collaborated with his colleague, Dr Patrick Tung from Research Technology Services, to explore the MONAI Framework. Their goal was to use the local computing capabilities and enable UNSW researchers to leverage AI acceleration in medical imaging via MONAI.

Patrick and Abdullah delved into two key components of MONAI:
1. MONAI Label: an AI-based image labelling and learning tool that applies reinforcement learning to train models.
2. MONAI Core: a tool that enables domain-specific capabilities like image transformation and segmentation.

The duo have set up the MONAI framework on Katana and have organised workshops to train and guide the wider research community in utilising this powerful tool on Katana. Additionally, they integrated the MONAI Label extension into an open-source image analysis software, 3D Slicer, accessible via Katana OnDemand. This integration allows the researchers to utilise MONAI’s AI capabilities within a user-friendly Graphical User Interface (GUI). 

According to Dr Patrick Tung
“MONAI is a game changer in Medical AI. Giving UNSW researchers the capability, platform and coaching on its use in their current workflows will help accelerate the research happening in medical imaging”.

A screenshot of a computer

Description automatically generated