Leveraging SO(3)-steerable convolutions for pose-robust semantic segmentation in 3D medical data
This work addresses pose robustness in medical image segmentation, offering a novel method that is domain-specific and incremental over standard CNNs.
The paper tackled the problem of pose sensitivity in 3D medical image segmentation by introducing SO(3)-steerable convolutional networks, resulting in improved robustness to unseen poses, enhanced segmentation performance in MRI brain tasks, and better parameter efficiency without requiring rotation-based data augmentation.
Convolutional neural networks (CNNs) allow for parameter sharing and translational equivariance by using convolutional kernels in their linear layers. By restricting these kernels to be SO(3)-steerable, CNNs can further improve parameter sharing. These rotationally-equivariant convolutional layers have several advantages over standard convolutional layers, including increased robustness to unseen poses, smaller network size, and improved sample efficiency. Despite this, most segmentation networks used in medical image analysis continue to rely on standard convolutional kernels. In this paper, we present a new family of segmentation networks that use equivariant voxel convolutions based on spherical harmonics. These networks are robust to data poses not seen during training, and do not require rotation-based data augmentation during training. In addition, we demonstrate improved segmentation performance in MRI brain tumor and healthy brain structure segmentation tasks, with enhanced robustness to reduced amounts of training data and improved parameter efficiency. Code to reproduce our results, and to implement the equivariant segmentation networks for other tasks is available at http://github.com/SCAN-NRAD/e3nn_Unet