Skip to main content
. Author manuscript; available in PMC: 2022 Jan 1.
Published in final edited form as: Med Image Anal. 2020 Oct 17;67:101880. doi: 10.1016/j.media.2020.101880

Fig. 4.

Fig. 4.

Consistency maps with the contours of the true kidney masks for three representative repetitions of the synthetic data experiment. The maps indicate the percentage of time that a voxel has been classified as kidney for each experiment. Voxels with values ∼ 100% indicate that these were consistently classified as kidney, while voxels with values between 0% and 100% shifted classification during the sequence. The masks registered with the different algorithms present good alignment with the ground truth masks. However, most algorithms present shifts in position of the mask in 10% of the time. LiMo-MoCo was the most consistent method in time while attaining similar accuracy values with the other competing methods (DICE coefficient 0.953).