Skip to main content
. Author manuscript; available in PMC: 2018 Nov 2.
Published in final edited form as: IEEE Trans Med Imaging. 2018 May 4;37(11):2428–2440. doi: 10.1109/TMI.2018.2833385

Fig. 3.

Fig. 3

The proposed 3D CNN classifier for A/V segmentation. 3D patches are extracted from the CT image around vessel candidates defined by a scale-space particle algorithm. The CNN learns A/V characteristics on these patches through three 3D convolutional layers, one max pooling and three fully connected layers.