Table 2.

Two-dimensional and three-dimensional DenseNet model structure details.

Convolutional layer	Two-dimensional DenseNet	Three-dimensional DenseNet
Pooling layer	7∗7	7∗7∗7
Compact block (1)	3∗3	3∗3∗3
Transition layer (1)	$\left|\begin{array}{c}1\ast 1\text{conv}\\ 3\ast 3\text{conv}\end{array}\right|\ast 6$	$\left|\begin{array}{c}1\ast 1\ast 1\text{conv}\\ 3\ast 3\ast 3\text{conv}\end{array}\right|\ast 6$
Compact block (2)	$\left|\begin{array}{c}1\ast 1\text{conv}\\ 2\ast 2\text{pool}\end{array}\right|$	$\left|\begin{array}{c}1\ast 1\ast 1\text{conv}\\ 2\ast 2\ast 2\text{pool}\end{array}\right|$
Transition layer (2)	$\left|\begin{array}{c}1\ast 1\text{conv}\\ 3\ast 3\text{conv}\end{array}\right|\ast 12$	$\left|\begin{array}{c}1\ast 1\ast 1\text{conv}\\ 3\ast 3\ast 3\text{conv}\end{array}\right|\ast 12$
Compact block (3)	$\left|\begin{array}{c}1\ast 1\text{conv}\\ 2\ast 2\text{pool}\end{array}\right|$	$\left|\begin{array}{c}1\ast 1\ast 1\text{conv}\\ 2\ast 2\ast 2\text{pool}\end{array}\right|$
Global pooling layer	$\left|\begin{array}{c}1\ast 1\text{conv}\\ 3\ast 3\text{conv}\end{array}\right|\ast 24$	$\left|\begin{array}{c}1\ast 1\ast 1\text{conv}\\ 3\ast 3\ast 3\text{conv}\end{array}\right|\ast 24$