Group Convolutional Neural Networks for DWI Segmentation

Research output: Contribution to journal › Conference article › Research

Standard

Group Convolutional Neural Networks for DWI Segmentation. / Liu, Renfei; Lauze, Francois Bernard; Bekkers, Erik J. ; Erleben, Kenny.

In: Proceedings of Machine Learning Research, Vol. 2022, No. 1, 2022, p. 1-11.

Research output: Contribution to journal › Conference article › Research

Harvard

Liu, R, Lauze, FB, Bekkers, EJ & Erleben, K 2022, 'Group Convolutional Neural Networks for DWI Segmentation', Proceedings of Machine Learning Research, vol. 2022, no. 1, pp. 1-11.

APA

Liu, R., Lauze, F. B., Bekkers, E. J., & Erleben, K. (2022). Group Convolutional Neural Networks for DWI Segmentation. Proceedings of Machine Learning Research, 2022(1), 1-11.

Vancouver

Liu R, Lauze FB, Bekkers EJ, Erleben K. Group Convolutional Neural Networks for DWI Segmentation. Proceedings of Machine Learning Research. 2022;2022(1):1-11.

Author

Liu, Renfei ; Lauze, Francois Bernard ; Bekkers, Erik J. ; Erleben, Kenny. / Group Convolutional Neural Networks for DWI Segmentation. In: Proceedings of Machine Learning Research. 2022 ; Vol. 2022, No. 1. pp. 1-11.

Bibtex

@inproceedings{c1bf943f142e4422a9373f3e587df2f5,
title = "Group Convolutional Neural Networks for DWI Segmentation",
abstract = "We present a Group Convolutional Network for Segmentation of Diffusion Weighted Imaging data (DWI). The network incorporates group actions that are natural for this type of data, in the form of equivariant convolutions, i.e., roto-translation equivariant convolutions. The equivariance property provides an important inductive bias and may alleviate the need for data augmentation strategies. Instead of performing group equivariant convolutions via spectral (Fourier-based) approaches, as is common for equivariance, we implement direct and light-weight regular group convolutions. We study the effect of equivariance and weight sharing over on performances of the networks on DWI scans from the Human Connectome project. We show how that full equivariance improves segmentations, while limiting the number of learnable parameters.",
author = "Renfei Liu and Lauze, {Francois Bernard} and Bekkers, {Erik J.} and Kenny Erleben",
year = "2022",
language = "English",
volume = "2022",
pages = "1--11",
journal = "Proceedings of Machine Learning Research",
issn = "2640-3498",
number = "1",
note = "GeoMedIA Workshop 2022 : Geometric Deep Learning in Medical Image Analysis ; Conference date: 18-11-2022",

}

RIS

TY - GEN

T1 - Group Convolutional Neural Networks for DWI Segmentation

AU - Liu, Renfei

AU - Lauze, Francois Bernard

AU - Bekkers, Erik J.

AU - Erleben, Kenny

PY - 2022

Y1 - 2022

N2 - We present a Group Convolutional Network for Segmentation of Diffusion Weighted Imaging data (DWI). The network incorporates group actions that are natural for this type of data, in the form of equivariant convolutions, i.e., roto-translation equivariant convolutions. The equivariance property provides an important inductive bias and may alleviate the need for data augmentation strategies. Instead of performing group equivariant convolutions via spectral (Fourier-based) approaches, as is common for equivariance, we implement direct and light-weight regular group convolutions. We study the effect of equivariance and weight sharing over on performances of the networks on DWI scans from the Human Connectome project. We show how that full equivariance improves segmentations, while limiting the number of learnable parameters.

AB - We present a Group Convolutional Network for Segmentation of Diffusion Weighted Imaging data (DWI). The network incorporates group actions that are natural for this type of data, in the form of equivariant convolutions, i.e., roto-translation equivariant convolutions. The equivariance property provides an important inductive bias and may alleviate the need for data augmentation strategies. Instead of performing group equivariant convolutions via spectral (Fourier-based) approaches, as is common for equivariance, we implement direct and light-weight regular group convolutions. We study the effect of equivariance and weight sharing over on performances of the networks on DWI scans from the Human Connectome project. We show how that full equivariance improves segmentations, while limiting the number of learnable parameters.

M3 - Conference article

VL - 2022

SP - 1

EP - 11

JO - Proceedings of Machine Learning Research

JF - Proceedings of Machine Learning Research

SN - 2640-3498

IS - 1

T2 - GeoMedIA Workshop 2022

Y2 - 18 November 2022

ER -

ID: 339166879