2024 Summer school on deep learning for medical imaging, 5th Edition

Welcome to the 2024 summer school on deep learning for medical imaging!

This school is intended for medical imaging beginners and experts (students, post-docs, research professionals, and professors) who are eager to discover the fundamentals of deep learning and how it translates to medical imaging. We will walk you through the basics of machine learning all the way to the latest deep learning breakthroughs applied to medical imaging. As shown in the planning, the school has both oral presentations (15 hours) and 4 hands-on sessions of 3 hours each. During the hands-on sessions, the participants will be guided through the dos and don'ts of machine learning programs for medical imaging. Please see the program for more details.

Should you be a machine learning / deep learning expert to attend the school? NO! That is the point of this school!

Should you be a programming expert to attend the hands-on session? NO! Only basic skills in Python programming are required.

We have enough room to accommodate 70 on-site participants. So do not wait and register!

Since we anticipate that more than 70 people will register to our summer school, we will grant the possibility for people to virtually attend the school in an asynchronous manner. These virtual participants will not be permitted on-site participation but will have access to the recorded videos as well as the hands-on session material towards the end of the school.

Please note that this school has been labeled by the MICCAI Society and the French National Scientific Research Counsil (MaDICS research group).

We would also like to thank our sponsors for their generous contributions.

Monday July 8th

8h00 - 9h00 Registration and Breakfast
9h00 - 9h15 Welcome talk
9h15 - 10h30 Basics of deep learning Part 1
Perceptron and multi-layer perceptron, stochastic gradient descent, learning rate, logistic regression, activation function, regularization (L1/L2/dropout/early stopping), etc.
10h30 - 10h45 Coffee Break
10h45 - 12h00 Basics of deep learning Part 2
Perceptron and multi-layer perceptron, stochastic gradient descent, learning rate, logistic regression, activation function, regularization (L1/L2/dropout/early stopping), etc.

12h00 - 13h30 Lunch at the ETS

13h30 - 15h00 Basics of deep learning Part 3
Weights initialization, forward and backward propagation, batch size, convolution neural nets (CNN), feature maps, pooling, pretraining and transfer learning, applications
15h00 - 15h30 Coffee break
15h30 - 17h00 Convolutional neural networks
Common CNN architectures for classification (VGGNet, ResNet, ...) and localization (FasterRCNN, Yolo) and segmentation (encoder-Decoder, U-Net, ENet, ...)

17h30 - 19h30 Cocktail at the ETS

Tuesday, July 9

8h30 - 9h00 Breakfast
9h00 - 10h30 Generative and adversarial methods for medical imaging
GANs, autoencoders, VAE and their application to medical imaging.
10h30 - 11h00 Coffee Break
11h00 - 12h30 Advanced concepts in deep learning Part 1
RNN, Attention, Transformers, ViT, etc.

12h30 - 14h00 Lunch at the ETS

15h00 - 15h30 Coffee break
14h00 - 16h30 Hands-on session 1: Introduction
Classification from machine learning to deep learning
16h30 - 18h00 Poster session

Wednesday, July 10

8h30 - 9h00 Breakfast
9h00 - 10h30 Uncertainty and explainability

Quality, uncertainty, calibration, explainability.

10h30 - 11h00 Coffee Break
11h00 - 12h30 Typical medical imaging issues
Domain adaptation, privacy protection and federated learning, adversarial learning, common pitfalls, incomplete data, etc.

12h30 - 14h00 Lunch at the ETS
14h00 - 17h00 Hands-on session 2: Segmentation using deep learning

Thursday, July 11

8h30 - 9h00 Breakfast
9h00 - 10h30 Weakly supervised deep learning
Weakly supervised segmentation, constrained CNN losses, semantic segmentation, semi-supervised learning
10h30 - 11h00 Coffee Break
11h00 - 12h30 Round table
What are the challenges for AI to break into clinic?

12h30 - 14h00 Lunch at the ETS
14h00 - 17h00 Hands-on session 3: Variational Autoencoders
Auto-encoders, convolutional auto-encoders, variational auto-encoders, latent spaces

18h00 - 21h30 Social event and banquet dinner

Friday, July 12

8h30 - 9h00 Breakfast
9h00 - 10h30 Advanced concepts in deep learning Part 2
Foundation models, text-image models, CLIP, diffusion networks, Dall-E, etc.
10h30 - 11h00 Coffee Break
11h00 - 12h30 MONAI : open-source, community-supported framework for Deep learning in healthcare imaging
Framework, Data Transformation, Common Network Definitions, Training Workflows, Tutorials, Model Zoo
12h30 - 12h45 Closing remarks

12h45 - 14h00 Lunch at the ETS
14h00 - 17h00 Hands-on session 4: Weakly supervised learning

Venue and Accommodation

Speakers

Ismail Ben Ayed

École de Technologie Supérieure, Canada

Dr. Ben Ayed received the PhD degree (with the highest honor) in computer vision from the INRS-EMT, Montreal in 2007. He is currently an Associate Professor at the École de Technologie Supérieure (ETS), University of Quebec. Before joining the ETS, he worked for 8 years as a research scientist at GE Healthcare, London, ON. He also holds an adjunct professor appointment at Western University (since 2012). Ismail's research interests include computer vision, optimization, machine learning and their potential applications in medical image analysis. He co-authored a book, over eighty peer-reviewed publications, mostly published in the top venues in these subject areas, and six US patents.
Pierre-Marc Jodoin

Université de Sherbrooke

Pierre-Marc Jodoin is from the University of Sherbrooke, Canada where he works as a full professor since 2007. He specializes in the development of novel techniques for machine learning and deep learning applied to computer vision and medical imaging. He mostly works in video analytics and brain and cardiac image analytics. He is the co-director of the Sherbrooke AI platform and co-founder of the medical imaging company called "Imeka.ca" which specializes in MRI brain image analytics. web site: http://info.usherbrooke.ca/pmjodoin/
Christian Desrosiers

École de technologie supérieure

Prof. Desrosiers obtained a Ph.D. in Applied Mathematics from Polytechnique Montreal in 2008, and was a postdoctoral researcher at the University of Minnesota with prof George Karypis. In 2009, he joined École de technologie supérieure (ÉTS) as professor in the Departement of Software and IT Engineering. He is codirector of the Laboratoire d’imagerie, de vision et d’intelligence artificielle (LIVIA) and a member of the REPARTI research network. He has over 100 publications in the fields of machine learning, image processing, computer vision and medical imaging, and has served on the scientific committee of several important conferences in these fields.
Jose Dolz

École de technologie supérieure, Canada

Jose Dolz is Associate Professor in the Department of Software and IT Engineering at the École de Technologie Supérieure (ETS) Montreal. Prior to be appointed Assistant Professor, he was a post-doctoral fellow at the ETS Montreal. Dr. Dolz obtained his B.Sc and M.Sc in the Polytechnic University of Valencia, Spain, and his Ph.D. at the University of Lille 2, France, in 2016. Dr. Dolz was recipient of a Marie-Curie FP7 Fellowship (2013-2016) to pursue his doctoral studies. His current research focuses on deep learning, medical imaging, optimization and learning strategies with limited supervision. He authored over 50 fully peer-reviewed papers, many of which published in the top venues in medical imaging (MedIA/TMI/MICCAI/IPMI), vision (CVPR) and learning (NeurIPS, ICML).
Michaël Sdika

University of Lyon, France

Michaël Sdika is a full member of the CREATIS lab in Lyon, France. His current research field focuses on the development of new analysis method based on deep learning for medical data. His main contributions are centered around image registration, atlas based segmentation, structure localization and machine learning for MR image of the nervous central system.
Thomas Grenier

University of Lyon, France

Dr. Thomas Grenier is Associate Professor at the University of Lyon Electrical Engineering department and at the CREATIS lab in Lyon, France.
His research focuses on longitudinal analysis of medical data to study evolution as Multiple Sclerosis lesions, functional activity (muscle and hydrocephaly). Most of these studies involve a segmentation task and dedicated pre and post processing steps. Clustering (spatio-temporal mean-shift), semi-supervised (multi-atlas with machine learning) or fully supervised (DNN) schemes are used to solve such problems considering their specific constraints.
Mohammadhadi Shateri

ETS

Mohammadhadi Shateri (Member, IEEE) received the B.Sc. degree (with Hons.) in electrical engineering from the Amirkabir University of Technology, Tehran, Iran, in 2012, the M.Sc. degree (with Hons.) in electrical engineering from the University of Manitoba, Winnipeg, MB, Canada, in 2017, and the Ph.D. in electrical engineering from McGill University, Montreal, QC, Canada, in 2021. He is now a professor at the École de Technologie Supérieure (ETS). His research interests include machine learning, deep learning, and reinforcement learning with application to data analytics.
Nicolas Thome

Sorbonne University (Paris)

His research interests include machine learning and deep learning for understanding low-level signals, e.g. vision, time series, acoustics, etc. He also explores solutions for combining low-level and higher-level data for multi-modal data processing. His current application domains are essentially targeted towards healthcare, physics and autonomous vehicles.
He is involved in several French, European and international collaborative research projects on artificial intelligence and deep learning.
Mohammad Havaei

Google

Mohammad Havaei is a Senior Research Scientist at Google. His research interest is focused on algorithmic fairness especially in the context of transfer learning. Mohammad's background is in the areas of computer vision, generative models and transfer learning. Before joining Google, he was a Research Scientist at Imagia where he focused on developing machine learning models applicable to medical image analysis.
Eric Kerfoot

King's College London

Dr Eric Kerfoot is a Software Architect in Medical Engineering in the Department of Biomedical Engineering, School of Biomedical Engineering & Imaging Sciences.