Comparison of atlas-based segmentation of subcortical structures in magnetic resonance brain images

This paper presents an atlas-based segmentation method for subcortical structures in magnetic resonance brain images. This method utilizes our previously introduced two-step registration comprising an affine transformation applied to the entire brain area followed by a local nonrigid transformation applied to subcortical structures. This method is compared to three existing atlas-based segmentation methods by using two objective segmentation indices, namely dice and relative error on volume. The IBSR database is considered for which expert segmented subcortical structures are available. The results obtained show that the proposed atlas-based segmentation method outperforms the existing atlas-based segmentation methods.