Domain Generalization for Mammography Detection via Multi-style and Multi-view Contrastive Learning

TL;DR

This paper introduces a contrastive learning approach that enhances mammography lesion detection across different vendors and image styles, especially with limited data, by learning invariant features through multi-style and multi-view training.

Contribution

It proposes a novel multi-style and multi-view contrastive learning scheme to improve domain generalization in mammography detection with limited vendor data.

Findings

Improves detection performance on unseen vendor data.

Outperforms existing state-of-the-art generalization methods.

Effective with limited training resources.

Abstract

Lesion detection is a fundamental problem in the computer-aided diagnosis scheme for mammography. The advance of deep learning techniques have made a remarkable progress for this task, provided that the training data are large and sufficiently diverse in terms of image style and quality. In particular, the diversity of image style may be majorly attributed to the vendor factor. However, the collection of mammograms from vendors as many as possible is very expensive and sometimes impractical for laboratory-scale studies. Accordingly, to further augment the generalization capability of deep learning model to various vendors with limited resources, a new contrastive learning scheme is developed. Specifically, the backbone network is firstly trained with a multi-style and multi-view unsupervised self-learning scheme for the embedding of invariant features to various vendor-styles. Afterward, the backbone network is then recalibrated to the downstream task of lesion detection with the specific supervised learning. The proposed method is evaluated with mammograms from four vendors and one unseen public dataset. The experimental results suggest that our approach can effectively improve detection performance on both seen and unseen domains, and outperforms many state-of-the-art (SOTA) generalization methods.