Multiatlas Segmentation Using Robust Feature-Based Registration

This paper presents a pipeline which uses a multi-atlas approach for multiorgan segmentation in whole-body CT images. In order to obtain accurate registrations between the target and the atlas images, we develop an adapted feature-based method which uses organ specific features. These features are learnt during an offline pre-processing step, and thus the algorithm still benefits from the speed of feature-based registration methods. These feature sets are then used to obtain pairwise non-rigid transformations using RANSAC, followed by a thin plate spline refinement or NIFTYREG. The fusion of the transferred atlas labels is performed using a random forest classifier, and finally the segmentation is obtained using graph cuts with a Potts model as interaction term. Our pipeline was evaluated on 20 organs in 10 whole-body CT images at the VISCERAL Anatomy Grand Challenge, in conjunction with the International Symposium on Biomedical Imaging, Brooklyn, New York, in April 2015. It performed best on a majority of the organs, with respect to the Dice index. Frida Fejne Department of Signals and Systems, Chalmers University of Technology, Sweden, e-mail: [email protected] Matilda Landgren Centre for Mathematical Sciences, Lund University, Sweden e-mail: [email protected] Jennifer Alvén Department of Signals and Systems, Chalmers University of Technology, Sweden, e-mail: [email protected] Johannes Ulén Centre for Mathematical Sciences, Lund University, Sweden e-mail: [email protected] Johan Fredriksson Centre for Mathematical Sciences, Lund University, Sweden e-mail: [email protected] Viktor Larsson Centre for Mathematical Sciences, Lund University, Sweden e-mail: [email protected] Olof Enqvist Department of Signals and Systems, Chalmers University of Technology, Sweden, e-mail: [email protected] Fredrik Kahl Department of Signals and Systems, Chalmers University of Technology, Sweden, Centre for Mathematical Sciences, Lund University, Sweden, e-mail: [email protected] ∗ The authors assert equal contribution and joint first authorship.