Skip to main content
. Author manuscript; available in PMC: 2022 Feb 1.
Published in final edited form as: Med Image Anal. 2020 Nov 5;68:101889. doi: 10.1016/j.media.2020.101889

Table 9:

Comparing the deeper encoder-decoder network structure SegCaps enabled by our proposed deconvolutional capsules, versus a network designed to be as similar as possible to CapsNet (Sabour et al., 2017) (Baseline SegCaps), abbreviated in table as Base-Caps.

Method Dice (%± std) HD (mm± std)
Base-Caps 75.97 ± 4.60 3 352.582 ± 133.451
SegCaps 88.92 ± 0.66 37.171 ± 23.223