The final type of segmentationmethod is atlas-based segmentation (sometimes also called label propagation). In this approach, additional knowledge is introduced through an atlas image, in which an expert has labeled the brain structures of interest. The atlas is first registered to the target image, and the resulting transformation is then used to deform the atlas labels to the coordinate system of the target image. During registration the similarity between the warped atlas image and the target image is maximized, while at the same time the deformation is constrained to ensure that the spatial information of the atlas is maintained.

MRI, brain images
W.J. Niessen (Wiro)
Erasmus University Rotterdam
Netherlands Organisation for Scientific Research (NWO), Erasmus MC
ASCI dissertation series
Erasmus MC: University Medical Center Rotterdam

van der Lijn, F. (2010, September 8). Automated Atlas-Based Segmentation of Brain Structures in MR Images: Application to a Population-Based Imaging Study (No. 215). ASCI dissertation series. Erasmus University Rotterdam. Retrieved from