Star Shape Prior in Fully Convolutional Networks for Skin Lesion Segmentation

Abstract

Semantic segmentation is an important preliminary step towards automatic medical image interpretation. Recently deep convolutional neural networks have become the first choice for the task of pixel-wise class prediction. While incorporating prior knowledge about the structure of target objects has proven effective in traditional energy-based segmentation approaches, there has not been a clear way for encoding prior knowledge into deep learning frameworks. In this work, we propose a new loss term that encodes the star shape prior into the loss function of an end-to-end trainable fully convolutional network (FCN) framework. We penalize non-star shape segments in FCN prediction maps to guarantee a global structure in segmentation results. Our experiments demonstrate the advantage of regularizing FCN parameters by the star shape prior and our results on the ISBI 2017 skin segmentation challenge data set achieve the first rank in the segmentation task among $21$ participating teams.

arXiv commentAccepted in MICCAI 2018

References in corpus (7)

• Multi-Scale Context Aggregation by Dilated Convolutions
• Anatomically Constrained Neural Networks (ACNN): Application to Cardiac Image Enhancement and Segmentation
• DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs
• Automatic Skin Lesion Analysis using Large-scale Dermoscopy Images and Deep Residual Networks
• ISIC 2017 - Skin Lesion Analysis Towards Melanoma Detection
• Understanding Convolution for Semantic Segmentation
• Incorporating prior knowledge in medical image segmentation: a survey

Cited by in corpus (2)

• Embracing Imperfect Datasets: A Review of Deep Learning Solutions for Medical Image Segmentation
• AUNet: Attention-guided dense-upsampling networks for breast mass segmentation in whole mammograms