Radiologist-Level Performance by Using Deep Learning for Segmentation of Breast Cancers on MRI Scans

Purpose To develop a deep network architecture that would achieve fully automated radiologist-level segmentation of cancers at breast MRI. Materials and Methods In this retrospective study, 38 229 examinations (composed of 64 063 individual breast scans from 14 475 patients) were performed in female patients (age range, 12–94 years; mean age, 52 years ± 10 [standard deviation]) who presented between 2002 and 2014 at a single clinical site. A total of 2555 breast cancers were selected that had been segmented on two-dimensional (2D) images by radiologists, as well as 60 108 benign breasts that served as examples of noncancerous tissue; all these were used for model training. For testing, an additional 250 breast cancers were segmented independently on 2D images by four radiologists. Authors selected among several three-dimensional (3D) deep convolutional neural network architectures, input modalities, and harmonization methods. The outcome measure was the Dice score for 2D segmentation, which was compared between the network and radiologists by using the Wilcoxon signed rank test and the two one-sided test procedure. Results The highest-performing network on the training set was a 3D U-Net with dynamic contrast-enhanced MRI as input and with intensity normalized for each examination. In the test set, the median Dice score of this network was 0.77 (interquartile range, 0.26). The performance of the network was equivalent to that of the radiologists (two one-sided test procedures with radiologist performance of 0.69–0.84 as equivalence bounds, P < .001 for both; n = 250). Conclusion When trained on a sufficiently large dataset, the developed 3D U-Net performed as well as fellowship-trained radiologists in detailed 2D segmentation of breast cancers at routine clinical MRI. Keywords: MRI, Breast, Segmentation, Supervised Learning, Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms Published under a CC BY 4.0 license. Supplemental material is available for this article.

[1]  David S. Melnick,et al.  International evaluation of an AI system for breast cancer screening , 2020, Nature.

[2]  Nan Wu,et al.  Deep Neural Networks Improve Radiologists’ Performance in Breast Cancer Screening , 2019, IEEE Transactions on Medical Imaging.

[3]  Jie Ding,et al.  Task‐based assessment of a convolutional neural network for segmenting breast lesions for radiomic analysis , 2019, Magnetic resonance in medicine.

[4]  Sidi Ahmed Mahmoudi,et al.  MRI Breast Tumor Segmentation Using Different Encoder and Decoder CNN Architectures , 2019, Comput..

[5]  Yu Huang,et al.  Segmentation of lesioned brain anatomy with deep volumetric neural networks and multiple spatial priors achieves human-level performance , 2019 .

[6]  Jun Zhang,et al.  Hierarchical Convolutional Neural Networks for Segmentation of Breast Tumors in MRI With Application to Radiogenomics , 2019, IEEE Transactions on Medical Imaging.

[7]  D. Bluemke,et al.  Multiparametric deep learning tissue signatures for a radiological biomarker of breast cancer: Preliminary results , 2018, Medical physics.

[8]  Jie Yang,et al.  A Spatio-Temporal Fully Convolutional Network for Breast Lesion Segmentation in DCE-MRI , 2018, ICONIP.

[9]  Karen Drukker,et al.  Most-enhancing tumor volume by MRI radiomics predicts recurrence-free survival “early on” in neoadjuvant treatment of breast cancer , 2018, Cancer Imaging.

[10]  Ritse Mann,et al.  Automated soft tissue lesion detection and segmentation in digital mammography using a u-net deep learning network , 2018, ArXiv.

[11]  István Csabai,et al.  Detecting and classifying lesions in mammograms with Deep Learning , 2017, Scientific Reports.

[12]  Christian Ledig,et al.  Checkerboard artifact free sub-pixel convolution: A note on sub-pixel convolution, resize convolution and convolution resize , 2017, ArXiv.

[13]  Gustavo Carneiro,et al.  Globally optimal breast mass segmentation from DCE-MRI using deep semantic segmentation as shape prior , 2017, 2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017).

[14]  Thomas Frauenfelder,et al.  Deep Learning in Mammography: Diagnostic Accuracy of a Multipurpose Image Analysis Software in the Detection of Breast Cancer , 2017, Investigative radiology.

[15]  Xiaohui Xie,et al.  Deep Multi-instance Networks with Sparse Label Assignment for Whole Mammogram Classification , 2016, bioRxiv.

[16]  Konstantinos Kamnitsas,et al.  Efficient multi‐scale 3D CNN with fully connected CRF for accurate brain lesion segmentation , 2016, Medical Image Anal..

[17]  Nico Karssemeijer,et al.  Large scale deep learning for computer aided detection of mammographic lesions , 2017, Medical Image Anal..

[18]  Gustavo Carneiro,et al.  The Automated Learning of Deep Features for Breast Mass Classification from Mammograms , 2016, MICCAI.

[19]  Mitchell D Schnall,et al.  Neoadjuvant Chemotherapy for Breast Cancer: Functional Tumor Volume by MR Imaging Predicts Recurrence-free Survival-Results from the ACRIN 6657/CALGB 150007 I-SPY 1 TRIAL. , 2016, Radiology.

[20]  Nico Karssemeijer,et al.  Unsupervised Deep Learning Applied to Breast Density Segmentation and Mammographic Risk Scoring , 2016, IEEE Transactions on Medical Imaging.

[21]  Gustavo Carneiro,et al.  Automated Mass Detection in Mammograms Using Cascaded Deep Learning and Random Forests , 2015, 2015 International Conference on Digital Image Computing: Techniques and Applications (DICTA).

[22]  Savannah C Partridge,et al.  Are Qualitative Assessments of Background Parenchymal Enhancement, Amount of Fibroglandular Tissue on MR Images, and Mammographic Density Associated with Breast Cancer Risk? , 2015, Radiology.

[23]  Ellen Warner,et al.  Effectiveness of screening with annual magnetic resonance imaging and mammography: results of the initial screen from the ontario high risk breast screening program. , 2014, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[24]  Les Irwig,et al.  Meta-analysis of magnetic resonance imaging in detecting residual breast cancer after neoadjuvant therapy. , 2013, Journal of the National Cancer Institute.

[25]  A. Nowacki,et al.  Understanding Equivalence and Noninferiority Testing , 2011, Journal of General Internal Medicine.

[26]  M. Giger,et al.  Cancerous breast lesions on dynamic contrast-enhanced MR images: computerized characterization for image-based prognostic markers. , 2010, Radiology.

[27]  Ying Lu,et al.  Kinetic assessment of breast tumors using high spatial resolution signal enhancement ratio (SER) imaging , 2007, Magnetic resonance in medicine.

[28]  C. Gatsonis,et al.  Cancer yield of mammography, MR, and US in high-risk women: prospective multi-institution breast cancer screening study. , 2007, Radiology.

[29]  M. Giger,et al.  A fuzzy c-means (FCM)-based approach for computerized segmentation of breast lesions in dynamic contrast-enhanced MR images. , 2006, Academic radiology.

[30]  Ellen Warner,et al.  Surveillance of BRCA1 and BRCA2 mutation carriers with magnetic resonance imaging, ultrasound, mammography, and clinical breast examination , 2004, JAMA.

[31]  Ron Kikinis,et al.  Statistical validation of image segmentation quality based on a spatial overlap index. , 2004, Academic radiology.

[32]  A. Vargha,et al.  A Critique and Improvement of the CL Common Language Effect Size Statistics of McGraw and Wong , 2000 .

[33]  Daniel Rueckert,et al.  Nonrigid registration using free-form deformations: application to breast MR images , 1999, IEEE Transactions on Medical Imaging.

[34]  J. Bezdek,et al.  FCM: The fuzzy c-means clustering algorithm , 1984 .

[35]  L. R. Dice.  Measures of the Amount of Ecologic Association Between Species , 1945 .