Kernel centered alignment supervised metric for Multi-Atlas segmentation

Abstract

Recently multi-atlas based methods have been used for supporting brain structure segmentation. These approaches encode the shape variability on a given population and provide prior information. However, the accuracy on the segmentation depend on the capability of the each atlas on the dataset to propagate the labels to the target image. In this sense, the selection of the most relevant atlases becomes an important task. In this paper, a new locally-weighted criterion is proposed to highlight spatial correspondences between images, aiming to enhance multi-atlas based segmentation results. Our proposal combines the spatial correspondences by a linear weighted combination and uses the kernel centered alignment criterion to find the best weight combination. The proposal is tested in an MRI segmentation task for state of the art image metrics as Mean Squares and Mutual Information and it is compared against other weighting criterion methods. Obtained results show that our approach outperforms the baseline methods providing a more suitable atlas selection and improving the segmentation of ganglia basal structures.