Mammography mass detection system based on deep learning in diagnosis of breast masses
10.13929/j.1003-3289.201908167
- Author:
Xingmei ZHANG
1
Author Information
1. Department of Radiology, United Family Healthcare
- Publication Type:Journal Article
- Keywords:
Artificial intelligence;
Breast neoplasms;
Deep learning;
Mammography
- From:
Chinese Journal of Medical Imaging Technology
2019;35(12):1794-1798
- CountryChina
- Language:Chinese
-
Abstract:
Objective: To observe the value of a mammogram mass detection system based on deep learning (DL) in diagnosis of breast masses. Methods: Data of 298 females who underwent mammography examination were retrospectively analyzed. The reference standards of mass detection were established by three senior radiologists. The lesion detection rate and detection stability of two radiologists with working time less than 5 years were compared and analyzed without (physician 1 and physician 2) or with artificial intelligence (AI) (physician 1+AI and physician 2+AI).Results: The lesion detection rate of physician 1+AI and physician 2+AI were all higher than that of physician 1 and physician 2 (both P<0.05). The detection rate of physician+AI was not affected by American college of radiology (ACR) breast density, breast imaging reporting and data system (BI-RADS), mass shape and mass density, etc (all P>0.05). Conclusion: The mammogram mass detection system based on DL can effectively improve mass detection rate of junior radiologists, and enhance the robustness of detection rate of different type masses.