Skin lesion segmentation using high-resolution convolutional neural network

Fengying Xie, Jiawen Yang, Jie Liu, Zhiguo Jiang, Yushan Zheng, Yukun Wang

Abstract: Highlights•A novel convolutional neural network is proposed, which can generate high-resolution feature maps to preserve spatial details.•The spatial and channel-wise attention mechanisms are adopted to enhance representative features while suppressing noise.•The proposed network can accurately extract skin lesion boundaries, and is robust to hair fibers and artifacts in the images.