Current browse context:
eess
Change to browse by:
References & Citations
Electrical Engineering and Systems Science > Image and Video Processing
Title: Dense Steerable Filter CNNs for Exploiting Rotational Symmetry in Histology Images
(Submitted on 6 Apr 2020 (v1), last revised 20 Jul 2020 (this version, v2))
Abstract: Histology images are inherently symmetric under rotation, where each orientation is equally as likely to appear. However, this rotational symmetry is not widely utilised as prior knowledge in modern Convolutional Neural Networks (CNNs), resulting in data hungry models that learn independent features at each orientation. Allowing CNNs to be rotation-equivariant removes the necessity to learn this set of transformations from the data and instead frees up model capacity, allowing more discriminative features to be learned. This reduction in the number of required parameters also reduces the risk of overfitting. In this paper, we propose Dense Steerable Filter CNNs (DSF-CNNs) that use group convolutions with multiple rotated copies of each filter in a densely connected framework. Each filter is defined as a linear combination of steerable basis filters, enabling exact rotation and decreasing the number of trainable parameters compared to standard filters. We also provide the first in-depth comparison of different rotation-equivariant CNNs for histology image analysis and demonstrate the advantage of encoding rotational symmetry into modern architectures. We show that DSF-CNNs achieve state-of-the-art performance, with significantly fewer parameters, when applied to three different tasks in the area of computational pathology: breast tumour classification, colon gland segmentation and multi-tissue nuclear segmentation.
Submission history
From: Simon Graham Mr [view email][v1] Mon, 6 Apr 2020 23:12:31 GMT (6783kb,D)
[v2] Mon, 20 Jul 2020 12:22:16 GMT (6862kb,D)
Link back to: arXiv, form interface, contact.