Affiliations: [a]
Qinghai Normal University, School of Physics and Electronic Information Engineering, Qinghai Province, Xining, China
| [b]
Lanzhou University, School of Information Science and Engineering, Gansu Province, Lanzhou, China
Correspondence:
[*]
Corresponding author. Yuelan Xin, Qinghai Normal University, School of Physics and Electronic Information Engineering, No. 38, Wusi West Road, Chengxi District, Xining City, Qinghai Province, Xining, 810008, China. Tel.: +86 13997282048; E-mail: [email protected].
Abstract: U-Net is a commonly used deep learning model for mammogram segmentation. Despite outstanding overall performance in segmenting, U-Net still faces from two aspects of challenges: (1) the skip-connections in U-Net have limitations, which may not be able to effectively extract multi-scale features for breast masses with diverse shapes and sizes. (2) U-Net only merges low-level spatial information and high-level semantic information through concatenating, which neglects interdependencies between channels. To address these two problems, we propose the U-shape adaptive scale network (ASU-Net), which contains two modules: adaptive scale module (ASM) and feature refinement module (FRM). In each level of skip-connections, ASM is used to adaptively adjust the receptive fields according to the different scales of the mass, which makes the network adaptively capture multi-scale features. Besides, FRM is employed to allows the decoder to capture channel-wise dependencies, which make the network can selectively emphasize the feature representation of useful channels. Two commonly used mammogram databases including the DDSM-BCRP database and the INbreast database are used to evaluate the segmentation performance of ASU-Net. Finally, ASU-Net obtains the Dice Index (DI) of 91.41% and 93.55% in the DDSM-BCRP database and the INbreast database, respectively.
