|Date:||Mon, November 16, 2009|
|Place:||East Hall 2|
Abstract: The segmentation of medical images like CT or MR is a great challenge due to the image noise. This image noise has an important impact on the result of a segmentation algorithm and thus on the shape and volume of the segmented object. A possibility to model the image noise is to perceive a voxel inside the image as a stochastic quantity, yielding to a stochastic image.
At the beginning of the talk I will present methods from classical image processing which I am investigating for stochastic extensions. After this introduction I will give a short overview of the field of stochastic partial differential equations (SPDE), which are needed for the segmentation on the stochastic images. For the discretization of the stochastic quantities in these SPDEs, I will introduce the Polynomial Chaos expansion. For the numerical treatment of the resulting SPDEs the MonteCarlo method and the Stochastic Finite Element method (SFEM) are presented. At the end of the talk a short overview over the problems arising for stochastic image segmentation and the planned progress of my PhD project is given. Also I will present first results of my investigations.