Images, segmentations and 3D models
A volumetric image, a segmentation layer and a spatial model serve different purposes. We explain how to compare them and what information should be retained so a result can be revisited.
Choosing a method
We discuss why input material, algorithm version and analysis settings matter. We explain why a result from a single example does not establish how a method will perform across all applications.
Evaluating and repeating an experiment
We describe ways to compare results and document computations. We also discuss failures and limitations that help define a method's scope of use.
Education and student projects
Learning image analysis through a defined experiment
We invite students and project supervisors to collaborate on tasks that combine medical imaging, programming and data analysis.
A starting point could be preparing an image for computation, comparing two segmentation methods, assessing the effect of parameters or presenting a result in 3D. Each task should have a defined objective, available material and evaluation criteria.
What the work involves
A project moves from a question, through method selection and an experiment, to a description of the results and limitations. Documentation should make it possible to return to the data and settings used.
For education, we use test material or other datasets approved for the project. Access and scope are agreed with the supervisor. Developing a method independently does not authorise its clinical use or make it available to other platform users.
Educational collaboration is agreed individually. This website does not offer open enrolment for a ready-made course.