Detection and classification of breast lesions using multiple information on contrast-enhanced mammography by a multiprocess deep-learning system: A multicenter study

• DOI: 10.21147/j.issn.1000-9604.2023.04.07
• Publication Date: 2023-09-07T08:32:15Z
• Authors: Yuqian Chen; ; Zhen Hua; Fan Lin; Tiantian Zheng; Heng Zhou; Shijie Zhang; Jing Gao; Zhongyi Wang; Huafei Shao; Wenjuan Li; Fengjie Liu; Simin Wang; Yan Zhang; Feng Zhao; Hao Liu; Haizhu Xie; Heng Ma; Haicheng Zhang; Ning Mao
• Keywords: Breast imaging; Feature (linguistics)

Accurate detection and classification of breast lesions in early stage is crucial to timely formulate effective treatments for patients. We aim develop a fully automatic system detect classify using multiple contrast-enhanced mammography (CEM) images.In this study, total 1,903 females who underwent CEM examination from three hospitals were enrolled as the training set, internal testing pooled external set prospective set. Here we developed CEM-based multiprocess (MDCS) perform task lesions. In system, introduced an innovative auxiliary feature fusion (AFF) algorithm that could intelligently incorporates types information images. The average free-response receiver operating characteristic score (AFROC-Score) was presented validate system's performance, performance evaluated by area under curve (AUC). Furthermore, assessed diagnostic value MDCS through visual analysis disputed cases, comparing its efficiency with radiologists exploring whether it augment radiologists' performance.On sets, always maintained high standalone AFROC-Scores 0.953 0.963 task, AUCs 0.909 [95% confidence interval (95% CI): 0.822-0.996] 0.912 CI: 0.840-0.985), respectively. It also achieved higher sensitivity than all senior specificity junior on sets. Moreover, performed superior reading time 5 seconds, compared 3.2 min. improved varying degrees assistance.MDCS demonstrated excellent lesions, greatly enhanced overall radiologists.