Adversarial Deep Structural Networks for Mammographic Mass Segmentation

TL;DR

This paper introduces an end-to-end deep learning framework combining fully convolutional networks and CRFs, enhanced with adversarial training, to improve mammographic mass segmentation accuracy on limited datasets.

Contribution

The novel integration of FCN, CRF, position priors, and adversarial training for mammogram segmentation is proposed, achieving state-of-the-art results.

Findings

Achieved superior segmentation accuracy on INbreast and DDSM-BCRP datasets.

Demonstrated effectiveness of adversarial training in small dataset scenarios.

Fused multiple models to enhance segmentation performance.

Abstract

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.