Rotation Equivariant CNNs for Digital Pathology
arXiv:1806.03962
Abstract
We propose a new model for digital pathology segmentation, based on the observation that histopathology images are inherently symmetric under rotation and reflection. Utilizing recent findings on rotation equivariant CNNs, the proposed model leverages these symmetries in a principled manner. We present a visual analysis showing improved stability on predictions, and demonstrate that exploiting rotation equivariance significantly improves tumor detection performance on a challenging lymph node metastases dataset. We further present a novel derived dataset to enable principled comparison of machine learning models, in combination with an initial benchmark. Through this dataset, the task of histopathology diagnosis becomes accessible as a challenging benchmark for fundamental machine learning research.
arXiv commentTo be presented at MICCAI 2018. Implementations of equivariant layers available at https://github.com/basveeling/keras_gcnn . PCam details and data at https://github.com/basveeling/pcam
References in corpus (6)
- Adam: A Method for Stochastic Optimization
- A Survey on Deep Learning in Medical Image Analysis
- Densely Connected Convolutional Networks
- Deep Learning for Identifying Metastatic Breast Cancer
- Detecting Cancer Metastases on Gigapixel Pathology Images
- Robustness of Rotation-Equivariant Networks to Adversarial Perturbations