Context-aware stacked convolutional neural networks for classification of breast carcinomas in whole-slide histopathology images

TLDR

Histopathological examination by a pathologist remains the gold standard for breast lesion diagnosis, yet automated whole‑slide image classification is difficult because benign lesions vary widely and ductal carcinoma in‑situ resembles invasive cancer at the cellular level. The study aims to enable high‑resolution tissue analysis with extensive contextual coverage to improve classification accuracy. The authors develop a context‑aware stacked CNN that first extracts cellular‑level features with a high‑resolution network, then feeds these into a second CNN trained on large patches to capture both fine‑grained details and global tissue structure, using 221 hematoxylin‑eosin stained WSIs. The system attains an AUC of 0.962 for binary non‑malignant versus malignant classification and an 81.3 % accuracy for distinguishing normal/benign, DCIS, and IDC, indicating strong potential for routine diagnostics.

Abstract

Currently, histopathological tissue examination by a pathologist represents the gold standard for breast lesion diagnostics. Automated classification of histopathological whole-slide images (WSIs) is challenging owing to the wide range of appearances of benign lesions and the visual similarity of ductal carcinoma in-situ (DCIS) to invasive lesions at the cellular level. Consequently, analysis of tissue at high resolutions with a large contextual area is necessary. We present context-aware stacked convolutional neural networks (CNN) for classification of breast WSIs into normal/benign, DCIS, and invasive ductal carcinoma (IDC). We first train a CNN using high pixel resolution to capture cellular level information. The feature responses generated by this model are then fed as input to a second CNN, stacked on top of the first. Training of this stacked architecture with large input patches enables learning of fine-grained (cellular) details and global tissue structures. Our system is trained and evaluated on a dataset containing 221 WSIs of hematoxylin and eosin stained breast tissue specimens. The system achieves an AUC of 0.962 for the binary classification of nonmalignant and malignant slides and obtains a three-class accuracy of 81.3% for classification of WSIs into normal/benign, DCIS, and IDC, demonstrating its potential for routine diagnostics.