Table 12 Summary of the HVU-ED segmenter model.

From: A novel hybrid vision UNet architecture for brain tumor segmentation and classification

Blocks

Layers

Input Size

Filter Size

No. of Filters

Activation Function

Output Size

 

Input

256\(\times\)256\(\times\)1

-

-

-

256\(\times\)256\(\times\)1

Encoder Block-1

Conv1

256\(\times\)256\(\times\)1

3\(\times\)3

64

ReLU

256\(\times\)256\(\times\)64

 

Conv2

256\(\times\)256\(\times\)64

3\(\times\)3

64

ReLU

256\(\times\)256\(\times\)64

 

MaxPooling

256\(\times\)256\(\times\)64

2\(\times\)2

-

-

128\(\times\)128\(\times\)128

Encoder Block-2

Conv1

128\(\times\)128\(\times\)64

3\(\times\)3

128

ReLU

128\(\times\)128\(\times\)128

 

Conv2

128\(\times\)128\(\times\)128

3\(\times\)3

128

ReLU

128\(\times\)128\(\times\)128

 

MaxPooling

128\(\times\)128\(\times\)128

2\(\times\)2

-

-

64\(\times\)64\(\times\)128

Encoder Block-3

Conv1

64\(\times\)64\(\times\)128

3\(\times\)3

256

ReLU

64\(\times\)64\(\times\)256

 

Conv2

64\(\times\)64\(\times\)256

3\(\times\)3

256

ReLU

64\(\times\)64\(\times\)256

 

MaxPooling

64\(\times\)64\(\times\)256

2\(\times\)2

-

-

32\(\times\)32\(\times\)256

Encoder Block-4

Conv1

32\(\times\)32\(\times\)256

3\(\times\)3

512

ReLU

32\(\times\)32\(\times\)512

 

Conv2

32\(\times\)32\(\times\)512

3\(\times\)3

512

ReLU

32\(\times\)32\(\times\)512

 

MaxPooling

32\(\times\)32\(\times\)512

2\(\times\)2

-

-

16\(\times\)16\(\times\)512

Encoder Block-5

Conv1

16\(\times\)16\(\times\)512

3\(\times\)3

1024

ReLU

16\(\times\)16\(\times\)1024

 

Conv2

16\(\times\)16\(\times\)1024

3\(\times\)3

1024

ReLU

16\(\times\)16\(\times\)1024

Bottleneck Block

ViT

256\(\times\)256\(\times\)3

-

-

GeLU

16\(\times\)16\(\times\)3

 

Hybrid Model

256\(\times\)256\(\times\)3

-

-

ReLU

16\(\times\)16\(\times\)3

 

UNet

16\(\times\)16\(\times\)1024

-

-

ReLU

16\(\times\)16\(\times\)1024

 

Concatenate

16\(\times\)16\(\times\)1024

-

-

-

32\(\times\)32\(\times\)1024

Decoder Block-1

Conv1

32\(\times\)32\(\times\)1024

3\(\times\)3

512

ReLU

32\(\times\)32\(\times\)512

 

Conv2

32\(\times\)32\(\times\)512

3\(\times\)3

512

ReLU

32\(\times\)32\(\times\)512

 

Concatenate

32\(\times\)32\(\times\)512

-

-

-

64\(\times\)64\(\times\)512

Decoder Block-2

Conv1

64\(\times\)64\(\times\)512

3\(\times\)3

256

ReLU

64\(\times\)64\(\times\)256

 

Conv2

64\(\times\)64\(\times\)256

3\(\times\)3

256

ReLU

64\(\times\)64\(\times\)256

 

Concatenate

64\(\times\)64\(\times\)256

-

-

-

128\(\times\)128\(\times\)256

Decoder Block-3

Conv1

128\(\times\)128\(\times\)512

3\(\times\)3

128

ReLU

128\(\times\)128\(\times\)128

 

Conv2

128\(\times\)128\(\times\)128

3\(\times\)3

128

ReLU

128\(\times\)128\(\times\)128

 

Concatenate

128\(\times\)128\(\times\)128

-

-

-

256\(\times\)256\(\times\)128

Decoder Block-4

Conv1

256\(\times\)256\(\times\)128

3\(\times\)3

64

ReLU

256\(\times\)256\(\times\)64

 

Conv2

256\(\times\)256\(\times\)64

3\(\times\)3

64

ReLU

256\(\times\)256\(\times\)64

Output

Conv3

256\(\times\)256\(\times\)64

1\(\times\)1

-

Softmax

256\(\times\)256\(\times\)1