Machine learning to improve brain cancer treatment.|
This project is currently retired, and not seeking new students.
Medical Image Processing Reading Group
The Brain Tumour Analysis Project is a collaboration between the University of Alberta's
Computing Science Department and
Cross Cancer Institute
to apply machine learning and computer vision techniques to the analysis of brain tumour patient MRI data.
The project's ultimate purpose is to apply these methods to improving the treatment of brain tumours (see
Project Description for details).
- Automatic tumour segmentation
- Brain tumour growth prediction
- Diagnosis and prognosis
- Database of brain tumour images
Methodologies / Techniques:
- Machine Learning
- Conditional random fields
- Bayesian belief networks
- Linear classifiers (eg: support vector machine, logistic regression)
- Computer Vision
- Level sets and variational methods
- Graph cuts
Outline of Navigation Panel (on the left)
- Home is this page.
- Project Description
gives a detailed description of the project, its goals, and relevant background.
summarizes our Automatic Segmentation Program, ASP.
- Growth Prediction
summarizes our overall approach to predicting how tumours will grow.
- Database summarizes our work on creating a database of patient scans as well as a
search tool to allow physicians and researchers to efficiently retrieve relevant cases from the database.
summarizes the software we have developed.
provides links to our publications and news releases about our work.
lists the people involved with this project.
- Sponsors lists supporting organizations.
has pointers to researchers not in our group who are working on similar projects along with relevant papers, software, etc.
Primer for some introductory material;
Publications for our publications and news releases about our work.)
- Glossary defines many of the technical terms
provides introductory information about the foundations of this project:
brains, tumours/cancer, MRI, imaging.
- Internal Use is accessible by project members only.
- Contact Us shows how to contact.