Medical Physics and Biomedical Engineering


Information Processing for Medical Imaging

Course Information

Unit value: 15 credits
Term: 2
Course organiser: Jamie McClelland
Other Lecturer: Yipeng Hu
Thursdays, 10am -1pm, Weeks 20-25, BBK 43 Gordon Square, Room 327 (Lecture theatre)
2pm - 5pm, Weeks 20-25, IOE 20 Bedford Way, Room 429 (Cluster room)


This module provides an essential introduction to theory and practice for information processing methods in medical imaging and computing. For students who have basic prior mathematics knowledge and programming experience, detailed concepts and examples of these “building block” topics are introduced. These, including medical image registration, segmentation and modern machine learning, are important computational components for understanding, developing and applying more advanced, applications-specific approaches in real-world medical imaging applications.

Aims and Objectives

The aim of Part A is to introduce medical image registration, one of the key tools used in modern medical image processing. The lectures will present the theoretical basis of medical image registration, including the essential building blocks of all image registration algorithms, and the details of some of the most widely used algorithms. They will also give real-world examples of the practical application of medical image registration, and the challenges and techniques for validating the results. The workshops will give hands-on experience with using and implementing medical image registration algorithms.

The objective for Part B is to introduce machine-learning-based methods for image segmentation, one of the most basic tasks in contemporary medical image computing, used for many medical imaging applications. Specifically, the tutorials and workshops are designed to guide the students to implement an example convolutional neural networks to segment anatomical structures from 3D volumetric images for diagnosis purposes. A basic understanding in parallel computing using Tensorflow is desirable but not required. The lectures cover an introduction to basic machine learning for regression and classification, while theory and practice in applying deep learning methods for supervised learning, such as the task set in the tutorials and workshops, are also introduced. This Part is designed to enable students a) to understand medical image segmentation problems and an overview of modern machine learning approaches; b) to obtain a working knowledge in deep learning based image segmentation methods; and c) to gain hands-on experience in developing and validating these methods.

Teaching and Exams

● Lectures, 3 hours x 6 lectures

● Lab session (tutorial/workshop), 3 hours x 6

● Un-assessed workshop problems sheets

● Private reading (papers, technical tutorials)

The assessment will consist of:

● Closed written examination (2 hours) 50%

● Assessed coursework, code and report 50%


Mathematics: basics in linear algebra (e.g. matrix operations and linear systems), basics in multivariate calculus (e.g. differentiation and the chain rule)

Programming: prior experience in Python and/or MATLAB


The course is designed for MSc students in the Department of Medical Physics and Biomedical Engineering and MRes students in CDT in Medical Imaging. In particular, this module provides an introduction of essential topics in medical image registration, segmentation and applied machine learning methods for those students who wish to take on future development or research involving computational methods in medical imaging. The course will be divided into lectures and lab sessions. Concepts and methodologies are detailed in the lectures and, in the lab sessions, real-world examples are provided in a mixed workshops and tutorials to expose students with practical experience.

Brief Syllabus

Part A. Medical Image Registration

  • What is medical image registration
  • The building blocks of a registration algorithm:

Cost function

  • Details of some popular registration algorithms
  • Example applications of medical image registration
  • Validating registration results

Part B. Machine Learning and Image Segmentation

● Probability theory and decision theory

● Regression and classification

● Topics in machine learning

● An introduction to deep learning

● Medical image segmentation methods

● Segmentation using convolutional neural networks

Core Texts

Part B. Machine Learning and Image Segmentation

● Bishop, C.M., (2006). Pattern Recognition and Machine Learning. Springer

● Goodfellow, I., Bengio, Y., Courville, A., (2016). Deep learning. MIT press. (http://www.deeplearningbook.org)