Interpretable Deep Neural Network to Predict Estrogen Receptor Status from Haematoxylin-Eosin Images
Standard
Interpretable Deep Neural Network to Predict Estrogen Receptor Status from Haematoxylin-Eosin Images. / Seegerer, Philipp; Binder, Alexander; Saitenmacher, René; Bockmayr, Michael; Alber, Maximilian; Jurmeister, Philipp; Klauschen, Frederick; Müller, Klaus Robert.
Artificial Intelligence and Machine Learning for Digital Pathology. Hrsg. / Andreas Holzinger; Randy Goebel; Michael Mengel; Heimo Müller. 1. Aufl. Cham : Springer Heidelberg, 2020. S. 16-37 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Band 12090 LNCS).Publikationen: SCORING: Beitrag in Buch/Sammelwerk › Kapitel › Forschung › Begutachtung
Harvard
APA
Vancouver
Bibtex
}
RIS
TY - CHAP
T1 - Interpretable Deep Neural Network to Predict Estrogen Receptor Status from Haematoxylin-Eosin Images
AU - Seegerer, Philipp
AU - Binder, Alexander
AU - Saitenmacher, René
AU - Bockmayr, Michael
AU - Alber, Maximilian
AU - Jurmeister, Philipp
AU - Klauschen, Frederick
AU - Müller, Klaus Robert
N1 - Publisher Copyright: © 2020, Springer Nature Switzerland AG. Copyright: Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2020
Y1 - 2020
N2 - The eligibility for hormone therapy to treat breast cancer largely depends on the tumor’s estrogen receptor (ER) status. Recent studies show that the ER status correlates with morphological features found in Haematoxylin-Eosin (HE) slides. Thus, HE analysis might be sufficient for patients for whom the classifier confidently predicts the ER status and thereby obviate the need for additional examination, such as immunohistochemical (IHC) staining. Several prior works are limited by either the use of engineered features, multi-stage models that use features unspecific to HE images or a lack of explainability. To address these limitations, this work proposes an end-to-end neural network ensemble that shows state-of-the-art performance. We demonstrate that the approach also translates to the prediction of the cancer grade. Moreover, subsets can be selected from the test data for which the model can detect a positive ER status with a precision of 94% while classifying 13% of the patients. To compensate for the reduced interpretability of the model that comes along with end-to-end training, this work applies Layer-wise Relevance Propagation (LRP) to determine the relevant parts of the images a posteriori, commonly visualized as a heatmap overlayed with the input image. We found that nuclear and stromal morphology and lymphocyte infiltration play an important role in the classification of the ER status. This demonstrates that interpretable machine learning can be a vital tool for validating and generating hypotheses about morphological biomarkers.
AB - The eligibility for hormone therapy to treat breast cancer largely depends on the tumor’s estrogen receptor (ER) status. Recent studies show that the ER status correlates with morphological features found in Haematoxylin-Eosin (HE) slides. Thus, HE analysis might be sufficient for patients for whom the classifier confidently predicts the ER status and thereby obviate the need for additional examination, such as immunohistochemical (IHC) staining. Several prior works are limited by either the use of engineered features, multi-stage models that use features unspecific to HE images or a lack of explainability. To address these limitations, this work proposes an end-to-end neural network ensemble that shows state-of-the-art performance. We demonstrate that the approach also translates to the prediction of the cancer grade. Moreover, subsets can be selected from the test data for which the model can detect a positive ER status with a precision of 94% while classifying 13% of the patients. To compensate for the reduced interpretability of the model that comes along with end-to-end training, this work applies Layer-wise Relevance Propagation (LRP) to determine the relevant parts of the images a posteriori, commonly visualized as a heatmap overlayed with the input image. We found that nuclear and stromal morphology and lymphocyte infiltration play an important role in the classification of the ER status. This demonstrates that interpretable machine learning can be a vital tool for validating and generating hypotheses about morphological biomarkers.
KW - Deep learning
KW - Digital pathology
KW - Explainable AI
U2 - 10.1007/978-3-030-50402-1_2
DO - 10.1007/978-3-030-50402-1_2
M3 - Chapter
AN - SCOPUS:85087544182
SN - 978-3-030-50401-4
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 16
EP - 37
BT - Artificial Intelligence and Machine Learning for Digital Pathology
A2 - Holzinger, Andreas
A2 - Goebel, Randy
A2 - Mengel, Michael
A2 - Müller, Heimo
PB - Springer Heidelberg
CY - Cham
ER -