Research on Classification Method of Medical Ultrasound Image Processing Based on Neural Network

Table 2

Structure of CNN4 used in this study.

Layer	Input	Filter	Padding	Stride	Output
Conv1	64 × 64, 2	5 × 5, 64	2 × 2	1 × 1	64 × 64, 64
Max-pooling1	64 × 64, 64	3 × 3	1 × 1	2 × 2	32 × 32, 64
Conv2	32 × 32, 64	5 × 5, 64	2 × 2	1 × 1	32 × 32, 64
Max-pooling2	32 × 32, 64	3 × 3	1 × 1	2 × 2	16 × 16, 64
Conv3	16 × 16, 64	3 × 3, 64	1 × 1	1 × 1	16 × 16, 64
Max-pooling3	16 × 16, 64	2 × 2	0 × 0	2 × 2	8 × 8, 64
Conv4	8 × 8, 64	3 × 3, 384	1 × 1	1 × 1	8 × 8, 384
Max-pooling4	8 × 8, 384	8 × 8	0 × 0	8 × 8	1 × 1, 384
Fc1	1 × 1, 384	—	—	—	1 × 1, 192
Fc2	1 × 1, 192	—	—	—	1 × 1, 1

conv denotes convolutional layer; Fc denotes fully connected layer; the format of the data in the table is size × number.