Skip to main content
. Author manuscript; available in PMC: 2019 Sep 5.
Published in final edited form as: IEEE Trans Med Imaging. 2019 Jan 23;38(9):2151–2164. doi: 10.1109/TMI.2019.2894322

Figure 5.

Figure 5

A block diagram illustrating how to explicitly introduce an anatomical shape refinement to the SSLLN-LR segmentation. As is evident in j, such a shape refinement enables an accurate, smooth and clinically meaningful bi-ventricular segmentation model, despite the artefacts in the LR input volume d. The framework is fully automated due to the use of the landmarks detected from SSLLN-HR and -LR.