Automated delineation of corneal layers on OCT images using a boundary-guided CNN

作者：

Highlights：

• A boundary-guided convolutional neural network (BG-CNN) was proposed to accurately and simultaneously segment different corneal layers and delineate their boundaries from OCT images.

• Two network modules were defined based on the classical U-Net network by introducing three different convolutional blocks.

• Experiment results on our collected OCT images demonstrated that the developed network achieved reasonable performance to identify corneal layers, as compared with several available networks.

摘要

•A boundary-guided convolutional neural network (BG-CNN) was proposed to accurately and simultaneously segment different corneal layers and delineate their boundaries from OCT images.•Two network modules were defined based on the classical U-Net network by introducing three different convolutional blocks.•Experiment results on our collected OCT images demonstrated that the developed network achieved reasonable performance to identify corneal layers, as compared with several available networks.

论文关键词：Corneal layers,OCT images,Segmentation,Convolutional neural networks

论文评审过程：Received 22 May 2020, Revised 24 May 2021, Accepted 3 July 2021, Available online 13 July 2021, Version of Record 15 July 2021.

论文官网地址：https://doi.org/10.1016/j.patcog.2021.108158