MLN-net: A multi-source medical image segmentation method for clustered microcalcifications using multiple layer normalization

Accurate segmentation of clustered microcalcifications in mammography iscrucial for the diagnosis and treatment of breast cancer. Despite exhibitingexpert-level accuracy, recent deep learning advancements in medical imagesegmentation provide insufficient contribution to practical applications, dueto the domain shift resulting from differences in patient postures, individualgland density, and imaging modalities of mammography etc. In this paper, anovel framework named MLN-net, which can accurately segment multi-source imagesusing only single source images, is proposed for clustered microcalcificationsegmentation. We first propose a source domain image augmentation method togenerate multi-source images, leading to improved generalization. And astructure of multiple layer normalization (LN) layers is used to construct thesegmentation network, which can be found efficient for clusteredmicrocalcification segmentation in different domains. Additionally, a branchselection strategy is designed for measuring the similarity of the sourcedomain data and the target domain data. To validate the proposed MLN-net,extensive analyses including ablation experiments are performed, comparison of12 baseline methods. Extensive experiments validate the effectiveness ofMLN-net in segmenting clustered microcalcifications from different domains andthe its segmentation accuracy surpasses state-of-the-art methods. Code will beavailable at https://github.com/yezanting/MLN-NET-VERSON1.