MICCAI Tutorial on AI for

Medical Image Analysis in Practice

👉 Registration

💁🏻‍♂ About Tutorial

Academia has made significant contributions to the advances in medical artificial intelligence(AI). Still, the industry is leading the state-of-the-art of AI models for real-world medical applications. Companies typically have better access to clinical data, dynamic annotations, and compute power when compared to academia. However, applying AI to real-world medical imaging problems comes with many additional challenges.

In this tutorial, we will give a broad overview of the entire process of building AI tools for real-world medical imaging problems in the industry. We will cover topics like data collection and intelligent annotation, model tuning and diagnosis for medical imaging problems specifically, AutoML and development infrastructure, evaluation of models from a clinical and regulatory perspective and finally filing for regulatory approval. These topics are of interest to academics or graduate students interested in practical applications of AI, or interested in transitioning to industry, and professionals that are new to the industry.

All speakers are researchers in industry and/or regulatory science with strong experience in applying AI to real-world medical problems. A keynote address will be provided by Prof. Dr. Michael Abramoff, the pioneer behind the first FDA-approved autonomous AI system.

▼ You can read the 'lecture note' here!

Tutorial on the development of AI models for medical image analysis (Thijs Kooi)

💡Topics

AI assisted annotation
Model tuning and diagnosis
AutoML research
Evaluation of AI models
Regulatory Strategy for AI Commercialization

💡Learning Objectives

Learning about annotations of medical image data and the real-world challenges that come with it.
Diagnosing an AI model for medical image data, identifying pain points of the model, and coming up with practical solutions to improve it.
Evaluation of the model with clinically relevant metrics.
Understanding principles and challenges of the regulatory assessment of AI-based software as a medical device
Real-world perspectives on ML research and development for MICCAI members seeking practical applications of AI, or a transition into the industry.

🗓 Program Schedule

🎤 Speakers

Subok Park, PhD

The Director of Clinical Strategy at Lunit.
Prior to joining Lunit in August of 2021, she worked as a regulatory scientist at the US Food and Drug Administration for over 16 years, gaining expertise in various capacities, such as pre-and post-market assessment of medical devices, objective assessment of AI/ML-based medical imaging devices, training and testing methodology of AI/ML CAD systems for disease detection and/or diagnosis, and balancing pre-and post-market study designs using simulated data and/or real-world data and evidence.

Contact: subok.park@lunit.io

Alexej Gossmann, PhD

A Staff Fellow at the FDA/CDRH/OSEL Division of Imaging, Diagnostics, and Software Reliability, where he splits his time evenly between research and regulatory work.
His research interest is in topics adjacent to performance evaluation of medical AI/ML, including test data reuse, generalization issues, imperfect reference standards, and other sources of bias in validation studies of AI/ML algorithms in medicine.
Alexej received his PhD degree from the interdisciplinary Bioinnovation Program at Tulane University, New Orleans, working in the areas of statistics, machine learning, genomics, and neuroimaging.
Alexej also holds a Bachelor of Science degree in mathematics from Technische Universität Darmstadt, Germany, and a Master's degree in statistics from Tulane University, New Orleans, where he also spent several years in the Mathematics PhD program.

Contact: linkedin.com/in/alexejgossmann

Minuk Ma

Leader of the Data-Centric AI team in the Oncology Research Department of Lunit.
He graduated with his MS degree from the Department of Electrical Engineering at Korea Advanced Institute of Science and Technology (KAIST) in 2020 where he studied deep neural network architectures for temporal moment localization in video using natural language query.
After joining Lunit in 2020, he engaged in developing novel computer vision based methods for discovering biomarkers for cancer treatment from pathology slide images.
He is particularly interested in developing methods for securing clean and correct data for AI model development.

Contact: minuk@lunit.io

Thijs Kooi, PhD

Vice President of research in the radiology group at Lunit.
Before joining Lunit he was at Vara, a German medical AI startup, where he built a triaging solution for screening mammography.
He got a M.Sc. in AI from the University of Amsterdam and a Ph.D. in computer aided diagnosis from Radboud University in the Netherlands.
His research interests are artificial intelligence and medical image analysis.

Contact: tkooi@lunit.io

Hyunjae Lee

Research Scientist at Lunit.
Working on deep learning with medical images.
He received his B.S. and M.S. degrees in Computer Science and Engineering from Seoul National University.
His research interests include computer vision and deep learning with a particular focus on AutoML algorithms in real-world scenarios.

Contact: hjlee@lunit.io

Click here to register! 🙌