Alert button

Adversarial Deep Structural Networks for Mammographic Mass Segmentation

Add code
Bookmark button
Alert button
Jun 09, 2017
Wentao Zhu, Xiang Xiang, Trac D. Tran, Xiaohui Xie

Figure 1 for Adversarial Deep Structural Networks for Mammographic Mass Segmentation
Figure 2 for Adversarial Deep Structural Networks for Mammographic Mass Segmentation
Figure 3 for Adversarial Deep Structural Networks for Mammographic Mass Segmentation
Figure 4 for Adversarial Deep Structural Networks for Mammographic Mass Segmentation

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon

Mass segmentation is an important task in mammogram analysis, providing effective morphological features and regions of interest (ROI) for mass detection and classification. Inspired by the success of using deep convolutional features for natural image analysis and conditional random fields (CRF) for structural learning, we propose an end-to-end network for mammographic mass segmentation. The network employs a fully convolutional network (FCN) to model potential function, followed by a CRF to perform structural learning. Because the mass distribution varies greatly with pixel position, the FCN is combined with position priori for the task. Due to the small size of mammogram datasets, we use adversarial training to control over-fitting. Four models with different convolutional kernels are further fused to improve the segmentation results. Experimental results on two public datasets, INbreast and DDSM-BCRP, show that our end-to-end network combined with adversarial training achieves the-state-of-the-art results.

* First version on arXiv 2016, MICCAI 2017 Deep Learning in Medical Image Analysis (DLMIA) workshop  
View paper onarxiv icon

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon