Hwang S, Park S. Accurate Lung Segmentation via Network-Wise Training of Convolutional Networks[M]//Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. Springer, Cham, 2017: 92-99.
1. Overview
In this paper, it proposed network-wise training
- 3 Conv + 6 Resblock (15 layers)
- dilation = 3 for the end of two residual blocks
- conv-BN
- global stride = 4
- fewer parameters
1.1. Details
- one fold (124 odd numbers) and one fold (123 even number)
- one fold train, another test, then calc avg
- resize to 256x256