Table 2.

Network structure of the U-Net architecture.

Module	Layer name	Kernel size	Stride	Kernel number	Output size
Input					128 × 128 × 4
	Conv 1	3 × 3	1	16	128 × 128 × 16
	Conv 2	3 × 3	1	16	128 × 128 × 16
	Max-pooling	2 × 2	2	16	64 × 64 × 16
	Conv 3	3 × 3	1	32	64 × 64 × 32
	Conv 4	3 × 3	1	32	64 × 64 × 32
	Max-pooling	2 × 2	2	32	32 × 32 × 32
	Conv 5	3 × 3	1	64	32 × 32 × 64
	Conv 6	3 × 3	1	64	32 × 32 × 64
Encoding	Max-pooling	2 × 2	2	64	16 × 16 × 64
	Conv 7	3 × 3	1	128	16 × 16 × 128
	Conv 8	3 × 3	1	128	16 × 16 × 128
	Max-pooling	2 × 2	2	128	8 × 8 × 128
	Conv 9	3 × 3	1	256	8 × 8 × 256
	Conv 10	3 × 3	1	256	8 × 8 × 256
	Conv-transpose	3 × 3	2	128	16 × 16 × 128
	Conv 11	3 × 3	1	128	16 × 16 × 128
Decoding	Conv 12	3 × 3	1	128	16 × 16 × 128
	Conv-transpose	3 × 3	2	64	32 × 32 × 64
	Conv 13	3 × 3	1	64	32 × 32 × 64
	Conv 14	3 × 3	1	64	32 × 32 × 64
	Conv-transpose	3 × 3	2	32	64 × 64 × 32
	Conv 15	3 × 3	1	32	64 × 64 × 32
	Conv 16	3 × 3	1	32	64 × 64 × 32
	Conv-transpose	3 × 3	2	16	128 × 128 × 16
	Conv 17	3 × 3	1	16	128 × 128 × 16
	Conv 18	3 × 3	1	16	128 × 128 × 16
Output	Conv 19	1 × 1	1	1	128 × 128 × 1