Application of contemporary image analysis and computer vision methods to medical imaging. Introduction into research developments in medical computer vision and image processing and solving of actual medical imaging problems through computer vision and processing systems.

Familiarising with basic terminology of medical image analysis and computer vision in medicine as well as basic numerical and image processing techniques useful in medical imaging such as multiresolution and multi-scale processing and optimisation. Practical application of learned image processing methods on real examples of digital radiography and computer vision methods on magnetic resonance images. Gaining basic understanding of contemporary computer vision algorithms used in medical imaging including image registration, segmentation as well as statistical modelling of anatomy shape and appearance. Each student will get at least two opportunities to apply learned techniques on actual medical images (project work and labs).

- Basics: digital medical images, 2D/3D, modalities, resolution, isotropy, dynamic images, temporal resolution, interpolation - Multiscale image analysis: analysis and sythesis, Gaussian and Laplacian pyramid, wavelets, DWT - Image processing for display: digital x-ray, range compression, image corrections, mutli-scale enhancement, noise suppression, tone scaling - Optimizacija: methods (gradient, simplex, LM...), distance measurement, hypothesis testing - Registration– image normalisation, perspective transformations, deformations, deformable registration, deformation fields, fluid registration, objective measures (MI, abs diff, sum sq) - Segmentation – illumination methods, snakes, level sets, mean shift, graf cuts, Markov random fields - Modeling of shape and appearance – statistical shape models, appearance models, texture and shape, active shape and apparance models (AAM)

The subject is delivered in three segments: - 12 double conventional lectures with electronic presentations - Laboratory exercises in Matlab environment (24 hours in 6 themes) - Individual student project focusing on a single imaging problem, in Matlab environment