Table 5.
CNN-Angio: the topology of the CNN for view classification in X-ray coronary angiographies.
| Layers | Parameters | Dimensions |
|---|---|---|
| Input | — | (1,256,256) |
| Convolution | (4,3,3) | (4,256,256) |
| Activation (sigmoid) | — | — |
| Average pooling | (2,2) | (4,128,128) |
| Convolution | (8,3,3) | (8,128,128) |
| Activation (tanh) | — | — |
| Average pooling | (2,2) | (8,64,64) |
| Convolution | (16,3,3) | (16,64,64) |
| Activation (tanh) | — | — |
| Average pooling | (2,2) | (16,32,32) |
| Convolution | (32,3,3) | (32,32,32) |
| Activation (tanh) | — | — |
| Average pooling | (2,2) | (32,16,16) |
| Flatten | — | (8192) |
| Fully connected | 64 | (64) |
| Activation (tanh) | — | — |
| Dropout | 25% | — |
| Fully connected | — | (1) |
| Activation (sigmoid) | — | — |