• Juan Zhou, Lu-Yang Luo, Qi Dou, Hao Chen, Cheng Chen, Gong-Jie Li, Ze-Fei Jiang, and Pheng-Ann Heng.
• Department of Radiology, The Fifth Medical Center of Chinese PLA General Hospital, Beijing, 10071, People's Republic of China.
• J Magn Reson Imaging. 2019 Oct 1; 50 (4): 1144-1151.

BackgroundThe usefulness of 3D deep learning-based classification of breast cancer and malignancy localization from MRI has been reported. This work can potentially be very useful in the clinical domain and aid radiologists in breast cancer diagnosis.PurposeTo evaluate the efficacy of 3D deep convolutional neural network (CNN) for diagnosing breast cancer and localizing the lesions at dynamic contrast enhanced (DCE) MRI data in a weakly supervised manner.Study TypeRetrospective study.SubjectsA total of 1537 female study cases (mean age 47.5 years ±11.8) were collected from March 2013 to December 2016. All the cases had labels of the pathology results as well as BI-RADS categories assessed by radiologists.Field Strength/Sequence1.5 T dynamic contrast-enhanced MRI.AssessmentDeep 3D densely connected networks were trained under image-level supervision to automatically classify the images and localize the lesions. The dataset was randomly divided into training (1073), validation (157), and testing (307) subsets.Statistical TestsAccuracy, sensitivity, specificity, area under receiver operating characteristic curve (ROC), and the McNemar test for breast cancer classification. Dice similarity for breast cancer localization.ResultsThe final algorithm performance for breast cancer diagnosis showed 83.7% (257 out of 307) accuracy (95% confidence interval [CI]: 79.1%, 87.4%), 90.8% (187 out of 206) sensitivity (95% CI: 80.6%, 94.1%), 69.3% (70 out of 101) specificity (95% CI: 59.7%, 77.5%), with the area under the curve ROC of 0.859. The weakly supervised cancer detection showed an overall Dice distance of 0.501 ± 0.274.Data Conclusion3D CNNs demonstrated high accuracy for diagnosing breast cancer. The weakly supervised learning method showed promise for localizing lesions in volumetric radiology images with only image-level labels.Level Of Evidence4 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2019;50:1144-1151.© 2019 International Society for Magnetic Resonance in Medicine.

26 keywords...

hide…