Generalizing Convolution Neural Networks on Stain Color Heterogeneous Data for Computational Pathology

Hematoxylin and Eosin (H&E) are one of the main tissue stains used in histopathology to discriminate between nuclei and extracellular material while performing a visual analysis of the tissue. However, histopathology slides are often characterized by stain color heterogeneity, due to different tissue preparation settings at different pathology institutes. Stain color heterogeneity poses challenges for machine learning-based computational analysis, increasing the difficulty of producing consistent diagnostic results and systems that generalize well. In other words, it is challenging for a deep learning architecture to generalize on stain color heterogeneous data, when the data are acquired at several centers, and particularly if test data are from a center not present in the training data. In this paper, several methods that deal with stain color heterogeneity are compared regarding their capability to solve center-dependent heterogeneity. Systematic and extensive experimentation is performed on a normal versus tumor tissue classification problem. Stain color normalization and augmentation procedures are used while training a convolutional neural networks (CNN) to generalize on unseen data from several centers. The performance is compared on an internal test set (test data from the same pathology institutes as the training set) and an external test set (test data from institutes not included in the training set). This also allows to measure generalization performance. An improved performance is observed when the predictions of the two best-performed stain color normalization methods with augmentation are aggregated. An average AUC and F1-score on external test are observed as 0.892 +/- 0.021 and 0.817 +/- 0.032 compared to the baseline 0.860 +/- 0.027 and 0.772 +/- 0.024 respectively.

Automated summary

The study compares methods for dealing with stain color heterogeneity in histopathology slides to improve machine learning-based computational analysis in normal versus tumor tissue classification. Through systematic experimentation, stain color normalization and augmentation techniques are used to train CNNs to generalize on unseen data from multiple centers, resulting in improved performance metrics on external test sets.