Se-DenseNet: Attention-Based Network for Detecting Pathological Images of Metastatic Breast Cancer

Clinically, the doctor judges the degree of tumor infiltration, and identifies whether it is metastasized by observing the histopathological section of the patient, and conducts targeted treatment for patients accordingly. Pathologists must be highly focused on the diagnostic process, but it is still prone to missed diagnosis. This paper presents a complete diagnostic system for the classification of pathological images of metastatic breast cancer, which can automatically identify metastatic breast cancer cells. In this paper, we use convolutional neural networks to extract images intensively and add attention mechanisms to score and re-learn the features to ensure that the effective features of metastatic cancer can be learned quickly and effectively. Furthermore, the generalization performance of the system is improved by using super-resolution to supplement image details. Eventually, the area under the corresponding curve (AUC) for a single model was 0.993, and the auc score on the kaggle test set was 0.001 higher than when the super-resolution treatment was not used. A final score of 0.975 was achieved on the KAGGLE Histopathologic Cancer Detection competition test set. This method can be used to assist doctors in examining pathological sections, and make the diagnosis results relatively objective.

[1]  Kilian Q. Weinberger,et al.  Densely Connected Convolutional Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[2]  Hao Chen,et al.  PFA-ScanNet: Pyramidal Feature Aggregation with Synergistic Learning for Breast Cancer Metastasis Analysis , 2019, MICCAI.

[3]  Aleksey Boyko,et al.  Detecting Cancer Metastases on Gigapixel Pathology Images , 2017, ArXiv.

[4]  LinLin Shen,et al.  Skin Lesion Analysis towards Melanoma Detection Using Deep Learning Network , 2017, Sensors.

[5]  Soumith Chintala,et al.  Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks , 2015, ICLR.

[6]  David Berthelot,et al.  BEGAN: Boundary Equilibrium Generative Adversarial Networks , 2017, ArXiv.

[7]  Takio Kurita,et al.  Fast and Accurate Image Super Resolution by Deep CNN with Skip Connection and Network in Network , 2017, ICONIP.

[8]  Xiaoou Tang,et al.  Learning a Deep Convolutional Network for Image Super-Resolution , 2014, ECCV.

[9]  Narendra Ahuja,et al.  Deep Laplacian Pyramid Networks for Fast and Accurate Super-Resolution , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[10]  Ross B. Girshick,et al.  Focal Loss for Dense Object Detection , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[11]  Aaron C. Courville,et al.  Improved Training of Wasserstein GANs , 2017, NIPS.

[12]  Kyoung Mu Lee,et al.  Accurate Image Super-Resolution Using Very Deep Convolutional Networks , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[13]  Hao Chen,et al.  ScanNet: A Fast and Dense Scanning Framework for Metastastic Breast Cancer Detection from Whole-Slide Image , 2017, 2018 IEEE Winter Conference on Applications of Computer Vision (WACV).

[14]  Nassir Navab,et al.  ReLayNet: Retinal Layer and Fluid Segmentation of Macular Optical Coherence Tomography using Fully Convolutional Network , 2017, Biomedical optics express.

[15]  I. Ellis,et al.  Pathological validation and significance of micrometastasis in sentinel nodes in primary breast cancer , 2001, Breast Cancer Research.

[16]  Kyunghyun Paeng,et al.  A Robust and Effective Approach Towards Accurate Metastasis Detection and pN-stage Classification in Breast Cancer , 2018, MICCAI.

[17]  Frank Hutter,et al.  SGDR: Stochastic Gradient Descent with Warm Restarts , 2016, ICLR.

[18]  Enhua Wu,et al.  Squeeze-and-Excitation Networks , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.