Skip to main content
. 2021 Oct 15;32(5):3131–3141. doi: 10.1007/s00330-021-08306-w

Fig. 2.

Fig. 2

Classification model architecture. Image patches from the detected tumour and corresponding position in the reference image were processed in two parallel pathways. The feature output from the two pathways was concatenated and processed by a series of fully connected layers and a final soft-max layer. The numbers shown in the image denotes the size of the feature dimension