All course materials were prepared by Mitko Veta and Maxime Lafarge.

Half-day version

Theory

• Machine learning fundamentals
• From linear models to deep neural networks
• Convolutional neural networks

Practice

• Training neural networks in your web browser using Tensorflow playground

Two-day version

Day 1

Theory

• Machine learning fundamentals
• From linear models to deep neural networks
• Convolutional neural networks

Practice

• Linear and logistic regression in Keras
• Fully connected neural networks for image classification in Keras
• Convolutional neural networks for image classification in Keras

Day 2

Theory

• Experimental methodology for training of (deep) machine learning models
• Modern neural network architectures

Practice

• Image segmentation with U-Net
• Mini-competition: segmentation of cardiac MR images (see part 2 of previous exercise)

Recommended resources

Some other resources that you might find useful after completing this course or as a preparation.

Other courses and materials from our group

• Educational module ‘Essential Skills for Machine Learning’
• 8P361 Project Imaging bachelor course
• 8DM40 Machine Learning in Medical Imaging and Biology master course

Other higly recommended online courses

• Oxford Machine Learning course by Nando de Freitas
• Stanford CS231n Convolutional Neural Networks for Visual Recognition course
• Stanford CS229 Machine Learning course

Books

• Elements of Statistical Learning
• Deep Learning
• Probabilistic Machine Learning: An Introduction

Review papers

• A survey on deep learning in medical image analysis
• High-performance medicine: the convergence of human and artificial intelligence

Datasets

A list of datasets published as medical image analysis challenges can be found here.

Videos

The Neural Networks video series from the 3Blue1Brown YoutubeChannel.