Skip to main content
. 2014 Oct 10;1(3):034003. doi: 10.1117/1.JMI.1.3.034003

Fig. 4.

Fig. 4

Architecture of the CNN model. The CNN architecture comprises three layers: two consecutive convolutional-pooling layers and a fully connected classification layer. The two convolutional-pooling layers use the same fixed 8×8 convolutional kernel and 2×2 pooling kernel, but have 64 and 128 neurons, respectively. The last layer has 256 neurons, which are all connected to the final two neurons for mitosis/nonmitosis classification.