From Detection of Individual Metastases to Classification of Lymph Node Status at the Patient Level: The CAMELYON17 Challenge

Bándi, Péter; Geessink, Oscar; Manson, Quirine F.; Dijk, Marcory van; Balkenhol, Maschenka; Hermsen, Meyke; Bejnordi, Babak Ehteshami; Lee, Byungjae; Paeng, Kyunghyun; Zhong, Aoxiao; Li, Quanzheng; Zanjani, Farhad Ghazvinian; Zinger, Svitlana; Fukuta, Keisuke; Komura, Daisuke; Ovtcharov, Vlado; Cheng, Shenghua; Zeng, Shaoqun; Thagaard, Jeppe; Dahl, Anders Bjorholm; Lin, Huangjing; Chen, Hao; Jacobsson, Ludwig; Hedlund, Martin; cetin, Melih; Halici, Eren; Jackson, Hunter; Chen, Richard; Both, Fabian; Franke, Jörg; Küsters‐Vandevelde, Heidi V.N.; Vreuls, Willem; Bult, Peter; Ginneken, Bram van; Laak, Jeroen van der; Litjens, Geert

doi:10.1109/tmi.2018.2867350

Cited by 403 publications

(326 citation statements)

References 24 publications

Supporting

Mentioning

322

Contrasting

Unclassified

Order By: Relevance

“…Most of the researches adopts a method of dividing exceptionally large WSIs into small patches and training classification models with patches. In particular, various methods using CNN were proposed in Camelyon Grand Challenge [3], a competition of lymph node metathesis classification. Lee et al [18], who achieved the highest score, also use a CNN as the classifier.…”

Section: Related Workmentioning

confidence: 99%

“…Camelyon16 dataset [4] and Camelyon17 dataset [3] are lymph node tissue slides for detecting breast cancer metastases. The maximum resolution is ×40.…”

Section: Datasetmentioning

confidence: 99%

“…Furthermore, based on the slide labels, the patientlevel metastasis label (pN-Stage) is defined. The assignment of the slide label and pN-stage follows the specified rule [3].…”

Section: Datasetmentioning

confidence: 99%

“…Next, we predict the slide-level label by training random forest as used in [18]. Finally, we predict the patient-level pN-stage based on the Camelyon17 metric [3] and evaluate with kappa score. We validate the performance with the validation data by 5-fold cross-validation setting.…”

Section: Evaluation On Camelyon17 Datasetmentioning

confidence: 99%

See 3 more Smart Citations

Multi-Stage Pathological Image Classification Using Semantic Segmentation

Takahama

Kurose

Mukuta

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

Histopathological image analysis is an essential process for the discovery of diseases such as cancer. However, it is challenging to train CNN on whole slide images (WSIs) of gigapixel resolution considering the available memory capacity. Most of the previous works divide high resolution WSIs into small image patches and separately input them into the model to classify it as a tumor or a normal tissue. However, patch-based classification uses only patch-scale local information but ignores the relationship between neighboring patches. If we consider the relationship of neighboring patches and global features, we can improve the classification performance. In this paper, we propose a new model structure combining the patch-based classification model and whole slide-scale segmentation model in order to improve the prediction performance of automatic pathological diagnosis. We extract patch features from the classification model and input them into the segmentation model to obtain a whole slide tumor probability heatmap. The classification model considers patch-scale local features, and the segmentation model can take global information into account. We also propose a new optimization method that retains gradient information and trains the model partially for end-to-end learning with limited GPU memory capacity. We apply our method to the tumor/normal prediction on WSIs and the classification performance is improved compared with the conventional patch-based method.

show abstract

Section: Related Workmentioning

confidence: 99%

“…Camelyon16 dataset [4] and Camelyon17 dataset [3] are lymph node tissue slides for detecting breast cancer metastases. The maximum resolution is ×40.…”

Section: Datasetmentioning

confidence: 99%

“…Furthermore, based on the slide labels, the patientlevel metastasis label (pN-Stage) is defined. The assignment of the slide label and pN-stage follows the specified rule [3].…”

Section: Datasetmentioning

confidence: 99%

Section: Evaluation On Camelyon17 Datasetmentioning

confidence: 99%

See 2 more Smart Citations

Multi-Stage Pathological Image Classification Using Semantic Segmentation

Takahama

Kurose

Mukuta

et al. 2019

2019 IEEE/CVF International Conference on Computer Vision (ICCV)

View full text Add to dashboard Cite

show abstract

“…To date, the bulk of machine learning approaches to digital pathology have focused on detection and segmentation of histologic primitives such as nuclear size and shape, grading of lesions (Campanella et al, 2019), prediction of clinical outcomes, and linking histopathology with other data types (Komura and Ishikawa, 2018;Madabhushi and Lee, 2016). Recent community challenges aimed at detecting lymph node metastases (CAMELYON16: Bandi et al, 2019;CAMELYON 17: Ehteshami Bejnordi et al, 2017), assessing tumor proliferation in breast cancer (TUPAC16: Veta et al, 2019), or detecting and classifying lung cancer (ACDC-LungHP, 2019). Other recent work has explored the prediction of genetic alterations from H&E-stained slides (Coudray et al, 2018;Kather et al, 2019).…”

Section: Introductionmentioning

confidence: 99%

A deep learning-based model of normal histology

Sing

Hoefling

Hossain

et al. 2019

Preprint

View full text Add to dashboard Cite

Deep learning models have been applied on various tissues in order to recognize malignancies. However, these models focus on relatively narrow tissue context or well-defined pathologies. Here, instead of focusing on pathologies, we introduce models characterizing the diversity of normal tissues. We obtained 1,690 slides with rat tissue samples from the control groups of six preclinical toxicology studies, on which tissue regions were outlined and annotated by pathologists into 46 different tissue classes. From these annotated regions, we sampled small patches of 224 x 224 pixels at six different levels of magnification. Using four studies as training set and two studies as test set, we trained VGG-16, ResNet-50, and Inception-v3 networks separately at each of these magnification levels. Among these models, Inception-v3 consistently outperformed the other networks and attained accuracies up to 83.4% (top-3 accuracy: 96.3%). Further analysis showed that most tissue confusions occurred within clusters of histologically similar tissues. Investigation of the embedding layer using the UMAP method revealed not only pronounced clusters corresponding to the individual tissues, but also subclusters corresponding to histologically meaningful structures that had neither been annotated nor trained for. This suggests that the histological representation learned by the normal histology network could also be used to flag abnormal tissue as outliers in the embedding space without a need to explicitly train for specific types of abnormalities. Finally, we found that models trained on rat tissues can be used on non-human primate and minipig tissues with minimal retraining. Significance statementLike many other scientific disciplines, histopathology has been profoundly impacted by recent advances in machine learning with deep neural networks. In this field, most deep learning models reported in the literature are trained on pathologies in specific tissues/contexts. Here, we aim to establish a model of normal tissues as a foundation for future models of histopathology. We build models that are specific to histopathology images and we show that their embeddings are better feature vectors for describing the underlying images than those of off-the shelf CNN models. Therefore, our models could be useful for transfer learning to improve the accuracy of other histopathology models.

show abstract