Shape classification and quantification in medical image analysis using the Riemannian geometry of spaces of tree-structured shapes 


The aim of the project is to create a mathematical framework for analysing treelike shapes in images, and in particular medical images - for instance in connection with vascular calcifications, smoker's lungs or cancer vascularization. In order to do statistics on images or shapes from images, we need to define a notion of distance between the shapes; in mathematical terms we want to define a geodesic metric structure on the space of shapes. Using the geometry of the metric we study deformations between treelike shapes and the corresponding distances between the shapes, enabling us to do classification of shapes with respect to diseases, which can be used for making medical diagnoses and prognoses.

Contact: Post doc Aasa Feragen