Table 1.

Densely connected encoder block of the proposed A-DenseUNet architecture. Note that “1 × 1, 192 conv” corresponds to 1 × 1 kernel size convolution with 192 features and a sequence of BN-Conv-ReLU layers. “[] × n” indicates n iterations of the dense block.

	Feature Size	Encoder DenseNet-164 (k = 48)
input	224 × 224 × 3	-
convolution 1	112 × 112	7 × 7, 96, stride 2
pooling	56 × 56	3 × 3 max pool, stride 2
dense block 1	56 × 56	$\left[\begin{array}{c}1 \times 1, 192 \mathrm{conv}\\ 3 \times 3, 48 \mathrm{conv}\end{array}\right]$ × 6
transition layer 1	56 × 56	1 × 1 conv
	28 × 28	2 × 2 average pool, stride 2
dense block 2	28 × 28	$\left[\begin{array}{c}1 \times 1, 192 \mathrm{conv}\\ 3 \times 3, 48 \mathrm{conv}\end{array}\right]$ × 12
transition layer 2	28 × 28	1 × 1 conv
	14 × 14	2 × 2 average pool, stride 2
dense block 3	14 × 14	$\left[\begin{array}{c}1 \times 1, 192 \mathrm{conv}\\ 3 \times 3, 48 \mathrm{conv}\end{array}\right]$ × 36
transition layer 3	14 × 14	1 × 1 conv
	7 × 7	2 × 2 average pool, stride 2
dense block 4	7 × 7	$\left[\begin{array}{c}1 \times 1, 192 \mathrm{conv}\\ 3 \times 3, 48 \mathrm{conv}\end{array}\right]$ × 24