Search papers, labs, and topics across Lattice.
This paper introduces rotation-equivariant convolutional neural networks into deformable brain MRI registration to improve alignment of anatomical structures. By replacing standard encoders with equivariant ones in existing architectures, the method achieves higher registration accuracy with fewer parameters, demonstrating the benefits of incorporating anatomical inductive biases. Experiments on multiple datasets show improved performance on rotated input pairs and with less training data, highlighting the robustness and sample efficiency of the approach.
Rotation-equivariant convolutions supercharge brain MRI registration, achieving higher accuracy with fewer parameters and greater robustness to orientation variations.
Image registration is a fundamental task that aligns anatomical structures between images. While CNNs perform well, they lack rotation equivariance - a rotated input does not produce a correspondingly rotated output. This hinders performance by failing to exploit the rotational symmetries inherent in anatomical structures, particularly in brain MRI. In this work, we integrate rotation-equivariant convolutions into deformable brain MRI registration networks. We evaluate this approach by replacing standard encoders with equivariant ones in three baseline architectures, testing on multiple public brain MRI datasets. Our experiments demonstrate that equivariant encoders have three key advantages: 1) They achieve higher registration accuracy while reducing network parameters, confirming the benefit of this anatomical inductive bias. 2) They outperform baselines on rotated input pairs, demonstrating robustness to orientation variations common in clinical practice. 3) They show improved performance with less training data, indicating greater sample efficiency. Our results demonstrate that incorporating geometric priors is a critical step toward building more robust, accurate, and efficient registration models.
