Table 4 Proposed BiLSTM CNN network for classifying CC mammogram images.
Layers | Types | Input size | Output size | Kernel size | Stride |
|---|---|---|---|---|---|
1 | Conv2D 1 | 224*224*3 | 150*150*3 | 7*7*3 | 1*1 |
2 | Conv2D 2 | 150*150*3 | 144*144*512 | 7*7*512 | 1*1 |
3 | Max Pooling 1 | 144*144*512 | 71*71*512 | 4*4 | 2*2 |
4 | Dropout 1 | 71*71*512 | 71*71*512 | Â | Â |
5 | Conv2D 3 | 71*71*512 | 65*65*256 | 7*7*256 | 1*1 |
6 | Max Pooling 2 | 65*65*256 | 31*31*256 | 4*4 | 2*2 |
7 | Dropout 2 | 31*31*256 | 31*31*256 | Â | Â |
8 | Conv2D 4 | 31*31*256 | 25*25*128 | 7*7*128 | 1*1 |
9 | Max Pooling 3 | 25*25*128 | 11*11*128 | 4*4 | 2*2 |
10 | Conv2D 5 | 11*11*128 | 7*7*64 | 5*5*64 | 1*1 |
11 | Max Pooing 4 | 7*7*64 | 2*2*64 | 4*4 | 2*2 |
12 | Reshape | 2*2*64 | 256*1 | Â | Â |
13 | BiLSTM | 256*1 | None,128 | Â | Â |
14 | Dropout 3 | 128*1 | 128*1 | Â | Â |
15 | Dense 1 | 128*1 | 128*1 | Â | Â |
16 | Dense 2 with output layer | 128*1 | 2*1 | Â | Â |
