Classification of Lung Nodules Based on Deep Residual Networks and Migration Learning

Table 3

Parameter configuration of the deep residual network, VGG19 model, and InceptionV3 model.

Deep residual network	VGG19	InceptionV3
Input: nodule/nonnodule images
conv1 7  7, 64	2  conv3-64	conv3-32
		conv3-32
		conv3-64
max pool	max pool	max pool
conv2_x	2  conv3-128	conv1-80
		conv3-192
	max pool	max pool
conv3_x	4  conv3-256	block1	module1  concat
			module2  concat
	max pool		module3  concat
conv4_x	4  conv3-512	block2	module1  concat
			module2  concat
			module3  concat
			module4  concat
	max pool		module5  concat
conv5_x	4  conv3-512	block3	module1  concat
			module2  concat
	max pool		module3  concat
Global average pooling2D
Fully connected layer-1024
Fully connected layer-2
Output: sigmoid