A global sudy of 2d and 3d statistical deformable models

Линиите свързват координатора с партньорите. За проекти отпреди 2014 г. CORDIS не винаги дава точни координати. Тези точки са на ниво град или държава.

Накратко на български

Цел на проекта

Research objectives and content One of the principal aims of image analysis in computer vision is shape modelling. The principal role of a model is to facilitate robust automatic interpretation even in noisy or cluttered images, or in images where parts of the objects of interest may be occluded. In the last few years, there has been a growing interest in deformable models for the analysis of images of objects whose appearance can vary continuously, in particular in the context of medical imaging, where organ shape may vary considerably through time and between individuals. Among several models which have been proposed, statistical deformable models seem to be more robust, because they are able to support a considerable shape variability but only deform in ways which are characteristic of the objects they represent. Their study is the topic of the proposed project. Specifically, the objectives of this research will be the following: further investigation into two existing methodologies, namely Cootes' and Grenander's deformable templates, in order to improve their efficiency; building of a common probabilistic framework for these two approaches; development of a new three-dimensional statistical model and finally, development of a tool based on the Bayesian paradigm for the fusion of different informations extracted from different modalities Because of the ability of statistical models to classify families of objects into subgroups according to shape variability and to identify abnormal cases, their application in the medical imaging field, especially with three-dimensional images, is very promising. Within the context of this project, we will particularly focus on the study of shape variability of the cardiac left ventricle from various modalities (Nuclear Medicine, Ultrasounds, MRI) in collaboration with the Department of Radiology of the University of Granada. Training content (objective, benefit and expected impact) The relevance of this research project is apparent in that it contributes to an understanding on how shape modelling may be better used in order to facilitate robust automatic interpretation of images containing structures of variable form. More specifically, this project aims to make an important contribution to the problem of defining deformable models which describe both typical shape and typical variability of objects, a relatively new endeavour in computer vision. In particular, the development of a three-dimensional statistical deformable model for studying shape variability in medical images will be pursued. The training of the applicant will be most benefited from the study of this type of deformable models which are characteristic of a specific class of objects. Links with industry / industrial relevance (22)

Оригинален текст от CORDIS (на английски).

Участници

• UNIVERSIDAD DE GRANADA · GRANADAКоординаторИспания
• Not availableНиво градФранция

Връзки

• Виж в CORDIS