Rotation Equivariant CNNs for Digital Pathology
Published in
International Conference on Medical Image Compu...(2018)External Links:
TL
TL;DR
This work proposes a new model for digital pathology segmentation, based on the observation that histopathology images are inherently symmetric under rotation and reflection, and presents a novel derived dataset to enable principled comparison of machine learning models, in combination with an initial benchmark.
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.
References17 items
1
Robustness of Rotation-Equivariant Networks to Adversarial Perturbations
2
Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer
3
Learning Steerable Filters for Rotation Equivariant CNNs
4
Supervised Intra-embedding of Fisher Vectors for Histopathology Image Classification
5