Deep Learning for Identifying Metastatic Breast Cancer
By Dayong Wang, Aditya Khosla, Rishab Gargeya, Humayun Irshad, Andrew H. Beck. 2016
1,005 citations recorded by the source corpus.
The International Symposium on Biomedical Imaging (ISBI) held a grand challenge to evaluate computational systems for the automated detection of metastatic breast cancer in whole slide images of sentinel lymph node biopsies. Our team won both competitions in the grand challenge, obtaining an area under the receiver operating curve (AUC) of 0.925 for the task of whole slide image classification and a score of 0.7051 for the tumor localization task. A pathologist independently reviewed the same images, obtaining a whole slide image classification AUC of 0.966 and a tumor localization score of 0.733. Combining our deep learning system's predictions with the human pathologist's diagnoses increased the pathologist's AUC to 0.995, representing an approximately 85 percent reduction in human error rate. These results demonstrate the power of using deep learning to produce significant improvements in the accuracy of pathological diagnoses.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Molecular Biology, Radiology, Nuclear Medicine and Imaging, Medical Imaging
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Yoshua Bengio
- Mohit Bansal
- Quoc V. Le
- Percy Liang
- Jianfeng Gao
- Sergey Levine
- Luke S. Zettlemoyer
- Ruslan Salakhutdinov
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Trevor Darrell
- Jitendra Malik
- Xiaogang Wang
- Ming-Hsuan Yang
- Andrew Zisserman
- Alan L. Yuille
In Molecular Biology
- Casey S. Greene
- Yifan Peng
- Brett K. Beaulieu-Jones
- G. Way
- Amr M. Alexandari
- Kyle Swanson
- Kevin W. Eliceiri
- Zhiyong Lu
In Radiology, Nuclear Medicine and Imaging
In Medical Imaging
- Fanjie Kong
- Ricardo Henao
- Anne L. Martel
- Ozan Ciga
- Chetan L. Srinidhi
- Francesco Ciompi
- J. A. van der Laak
- Geert J. S. Litjens
Citations in the corpus
- Rotation Equivariant CNNs for Digital Pathology (2018, background): “4 reports the performance on the FROC score, next to those of a pathologist [9] and the state-of-the-art approaches reported on this dataset, including [1,17].”
- Improving the accuracy of medical diagnosis with causal machine learning (2020, methodology): “…and using Bayes rule to estimate the posterior, P ð D jE ; θ Þ ¼ P ðEj D ; θ Þ P ð D ; θ Þ P ðE ; θ Þ : Examples of discriminative diagnostic models include neural network and deep learning models 8,10,15,33,34 , whereas generative models are typically Bayesian networks 18,19,21,22,27,35,36 .”
- Opportunities and obstacles for deep learning in biology and medicine (2017, background): “Cooperation between human experts and deep learning algorithms addresses many of these challenges and can achieve better performance than either individually [66].”
- Detecting Cancer Metastases on Gigapixel Pathology Images (2017, methodology): “By contrast, some existing methods pre-sample a set of patches from each slide [23], which limits the breadth of patches seen during training.”
- Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017: 20th International Conference, Quebec City, QC, Canada, September 11-13, 2017, Proceedings, Part II (2017, positive): “In instance weighting, the training samples are assigned weights according to the distribution of the labels (class imbalance) [10] and the distribution of the observations (covariate shift) [16].”
- Artificial Intelligence: The Ambiguous Labor Market Impact of Automating Prediction (2019, methodology): “This approach uses artificial intelligence to augment the diagnosis decisions of humans rather than replace them altogether (Wang et al 2016).”
- Transfusion: Understanding Transfer Learning with Applications to Medical Imaging (2019, neutral): “On the one hand, transfer learning from ImageNet has been almost universally adopted Shin et al. (2016); Rajpurkar et al. (2017); Wang et al. (2016); Gulshan et al. (2016); Liu et al. (2017); Raghu et al. (2018), as models must learn to process large ∗Equal Contribution †Equal Contribution ar X iv…”
- Efficient Classification of Very Large Images with Tiny Objects (2021, methodology): “Incorporating pixel-level annotations via pre-training A good attention function and a feature extractor can be obtained by training the model components as tile-level classifiers [21, 22].”
- Key challenges for delivering clinical impact with artificial intelligence (2019, background): “It has also been shown that humans assisted by AI performed better than either alone in a study of diabetic retinopathy screening [96, 97].”
- Activation Functions: Comparison of trends in Practice and Research for Deep Learning (2018, background): “recognition [27], [28], segmentation [29], [30], scene understanding and description [31], [32], machine translation [33], [34] test to speech systems [35], [36], cancer detection systems [37], [38], [39], finger print detection [40], [41], weather forecast [42], [43], self-driving cars [44], [45], ”
- Predicting cancer outcomes from histology and genomics using convolutional networks (2017, methodology): “Aided by advances in slide scanning microscopes and computing, a number of image analysis algorithms have been developed for grading (1 – 4), classification (5 – 10), and identification of lymph node metastases (11) in multiple cancer types.”
- Transfusion: Understanding Transfer Learning for Medical Imaging (2019, background): “n. This question is especially relevant in the context of medical imaging. On the one hand, transfer learning from ImageNet has been almost universally adoptedShin et al.(2016);Rajpurkar et al.(2017);Wang et al. (2016);Gulshan et al.(2016);Liu et al.(2017);Raghu et al.(2018), as models must learn to”
- Deep Learning for Whole Slide Image Analysis: An Overview (2019, background): “Most successful approaches also employ hard negative mining, an iterative process whereby false positives are added to the training data set for further training (14, 23, 24, 36).”
- Deep learning in histopathology: the path to the clinic (2021, background): “Current applications include tumor detection and classification (often by sub-type 23,25–39 ), image segmentation 40–50 , cell detection and counting 51–55 , mitosis detection 56–60 , analysis of kidney transplant biopsies 20 and tumor grading 61–63 among others.”