Spatial and Planar Consistency for Semi-Supervised Volumetric Medical Image Segmentation
Yanfeng Zhou (Institute of Automation, Chinese Academy of Sciences), yiming huang (nstitute of Automation，Chinese Academy of Sciences), Ge Yang (National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences)*
The 34th British Machine Vision Conference
Abstract
Semi-supervised volumetric medical image segmentation have achieved remarkable success with the development of deep neural networks (DNNs). Consistency regularization is a common strategy for semi-supervised volumetric segmentation. These models use perturbations (such as noise, distance mapping, dropout, etc.) to construct consistency losses. So far, however, few studies exploit the differences of spatial and planar information between 2D and 3D models for consistency learning. In this study, we propose a spatial and planar consistency (SPC) strategy, which outperforms previous state-of-the-art models in semi-supervised volumetric medical image segmentation. SPC consists of a 3D spatial branch and a 2D planar branch. The 3D spatial branch focuses on the complete spatial structure of segmentation objects, while the 2D planar branch focuses on the planar details. The outputs of two branches can be used for semi-supervised consistency learning. Extensive experiments on two public 3D datasets demonstrate the effectiveness of our model. Code is available at https://github.com/Yanfeng-Zhou/SPC.Citation
@inproceedings{Zhou_2023_BMVC,
author = {Yanfeng Zhou and yiming huang and Ge Yang},
title = {Spatial and Planar Consistency for Semi-Supervised Volumetric Medical Image Segmentation},
booktitle = {34th British Machine Vision Conference 2023, {BMVC} 2023, Aberdeen, UK, November 20-24, 2023},
publisher = {BMVA},
year = {2023},
url = {https://bmvc2022.mpi-inf.mpg.de/BMVC2023/0084.pdf}
}
Copyright © 2023
The British Machine Vision Conference is organised by