||Department of Oncology, Radiology and Clinical Immunology, UU
||Atlas-based segmentation of the prostate for the planning of radiotherapy
||In radiotherapy targets as well as organs at risk (OARs) are manually segmented in clinical routine for creation and evaluation of treatment plans. Recent hardware improvements such as 4D-CT scanning and the possibility to acquire cone beam CT scans directly at the treatment unit will add to this segmentation burden. As the CBA audience is undoubtedly aware of, manual segmentation is a tedious process with a high degree of inter- and intra-user variability.
I will talk about atlas-based segmentation, where we transfer information from previously segmented patients to construct a new segmentation by the use of deformable registration. We register several old patient images to one new image set and use similarity measures to merge the resulting segmentation proposals to a single segmentation. I will present results for a prostate segmentation using a database of 22 patients.