Table 1 Presents the architectural components of the proposed SWNet model

Input layer

224×224×3

Conv1, BN1, ReLU1

Conv1: Kernel size=3×3

224×224×32

Conv2, BN2, ReLU2

Conv2: Kernel size=3×3, stride=1

224×224×32

Conv3, BN3, ReLU3

Conv3: Kernel size=3×3, stride=2

112×112×32

Conv4, BN4, ReLU4

Conv4: Kernel size=3×3, stride=1

224×224×32

Conv5, BN5, ReLU5

Conv5: Kernel size=3×3, stride=2

112×112×32

Conv6, BN6, ReLU6

Conv6: Kernel size=3×3, stride=1

224×224×32

Conv7, BN7, ReLU7

Conv7: Kernel size=3×3, stride=2

112×112×32

Conv8, BN8, ReLU8

Conv8: Kernel size=3×3, stride=1

224×224×32

Conv9, BN9, ReLU9

Conv9: Kernel size=3×3, stride=2

112×112×32

Concat1, BNConcat1

Concatenation of four inputs

112×112×128

Conv10,BN10,ReLU10

Conv10: Kernel size=3×3, stride=1

112×112×64

Conv11, BN11, ReLU11

Conv11: Kernel size=3×3, stride=2

56×56×64

Conv12, BN12, ReLU12

Conv12: Kernel size=3×3, stride=1

112×112×64

Conv13, BN13, ReLU13

Conv13: Kernel size=3×3, stride=2

56×56×64

Conv14, BN14, ReLU14

Conv14: Kernel size=3×3, stride=1

112×112×64

Conv15, BN15, ReLU15

Conv15: Kernel size=3×3, stride=2

56×56×64

Conv16, BN16, ReLU16

Conv16: Kernel size=3×3, stride=1

112×112×64

Conv17, BN17, ReLU17

Conv17: Kernel size=3×3, stride=2

56×56×64

Concat2, BNConcat2

Concatenation of four inputs

56×56×64

Conv18, BN18, ReLU18

Conv18: Kernel size=3×3, stride=1

56×56×64

Conv19, BN19, ReLU19

Conv19: Kernel size=3×3, stride=2

28×28×128

Conv20, BN20, ReLU20

Conv20: Kernel size=3×3, stride=1

56×56×128

Conv21, BN21, ReLU21

Conv21: Kernel size=3×3, stride=2

28×28×128

Conv22, BN22, ReLU22

Conv22: Kernel size=3×3, stride=1

56×56×128

Conv23, BN23, ReLU23

Conv23: Kernel size=3×3, stride=2

28×28×128

Conv24, BN24, ReLU24

Conv24: Kernel size=3×3, stride=1

56×56×128

Conv25, BN25, ReLU25

Conv25: Kernel size=3×3, stride=2

28×28×128

Concat3, BNConcat3

Concatenation of four inputs

28×28×512

Conv26, BN26, ReLU26

Conv26: Kernel size=3×3, stride=1

28×28×256

Conv27, BN27, ReLU27

Conv27: Kernel size=3×3, stride=2

14×14×256

Conv28, BN28, ReLU28

Conv28: Kernel size=3×3, stride=1

28×28×256

Conv29, BN29, ReLU29

Conv29: Kernel size=3×3, stride=2

14×14×256

Conv30, BN30, ReLU30

Conv30: Kernel size=3×3, stride=1

28×28×256

Conv31, BN31, ReLU31

Conv31: Kernel size=3×3, stride=2

14×14×256

Conv32, BN32, ReLU32

Conv32: Kernel size=3×3, stride=1

28×28×256

Conv33, BN33, ReLU33

Conv33: Kernel size=3×3, stride=2

14×14×256

Concat4, BNConcat4

Concatenation of four inputs

14×14×1024

Average polling layer

1×1×1024

Fc1

300 fully connected

1×1×300

Drop1

Dropout layer with learning rate:0.5

1×1×300

Fc2

64 fully connected

1×1×64

Drop2

Dropout layer with learning rate:0.5

1×1×300

Fc3 (Softmax layer)

0= Benign, 1= Malignant

1×1×2