Voxel deconvolutional networks for 3D brain image labeling

Yongjun Chen, Min Shi, Hongyang Gao, Dinggang Shen, Lei Cai, Shuiwang Ji

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

Deep learning methods have shown great success in pixel-wise prediction tasks. One of the most popular methods employs an encoder-decoder network in which deconvolutional layers are used for up-sampling feature maps. However, a key limitation of the deconvolutional layer is that it suffers from the checkerboard artifact problem, which harms the prediction accuracy. This is caused by the independency among adjacent pixels on the output feature maps. Previous work only solved the checkerboard artifact issue of deconvolutional layers in the 2D space. Since the number of intermediate feature maps needed to generate a deconvolutional layer grows exponentially with dimensionality, it is more challenging to solve this issue in higher dimensions. In this work, we propose the voxel deconvolutional layer (VoxelDCL) to solve the checkerboard artifact problem of deconvolutional layers in 3D space. We also provide an efficient approach to implement VoxelDCL. To demonstrate the effectiveness of VoxelDCL, we build four variations of voxel deconvolutional networks (VoxelDCN) based on the U-Net architecture with VoxelDCL. We apply our networks to address volumetric brain images labeling tasks using the ADNI and Loni LPBA40 datasets. The experimental results show that the proposed iVoxelDCNa achieves improved performance in all experiments. It reaches 83.34% in terms of dice ratio on the ADNI dataset and 79.12% on the Loni LPBA40 dataset, which increases 1.39% and 2.21% respectively compared with the baseline. In addition, all the variations of VoxelDCN we proposed outperform the baseline methods on the above datasets, which demonstrates the effectiveness of our methods.

Original languageEnglish
Title of host publicationKDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
PublisherAssociation for Computing Machinery
Pages1226-1234
Number of pages9
ISBN (Print)9781450355520
DOIs
Publication statusPublished - 2018 Jul 19
Externally publishedYes
Event24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2018 - London, United Kingdom
Duration: 2018 Aug 192018 Aug 23

Other

Other24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2018
CountryUnited Kingdom
CityLondon
Period18/8/1918/8/23

Fingerprint

Labeling
Brain
Pixels
Sampling
Experiments
Deep learning

Keywords

  • Deep learning
  • Volumetric brain image labeling
  • Voxel deconvolutional layer
  • Voxel deconvolutional networks

ASJC Scopus subject areas

  • Software
  • Information Systems

Cite this

Chen, Y., Shi, M., Gao, H., Shen, D., Cai, L., & Ji, S. (2018). Voxel deconvolutional networks for 3D brain image labeling. In KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 1226-1234). Association for Computing Machinery. https://doi.org/10.1145/3219819.3219974

Voxel deconvolutional networks for 3D brain image labeling. / Chen, Yongjun; Shi, Min; Gao, Hongyang; Shen, Dinggang; Cai, Lei; Ji, Shuiwang.

KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery, 2018. p. 1226-1234.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Chen, Y, Shi, M, Gao, H, Shen, D, Cai, L & Ji, S 2018, Voxel deconvolutional networks for 3D brain image labeling. in KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery, pp. 1226-1234, 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2018, London, United Kingdom, 18/8/19. https://doi.org/10.1145/3219819.3219974
Chen Y, Shi M, Gao H, Shen D, Cai L, Ji S. Voxel deconvolutional networks for 3D brain image labeling. In KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery. 2018. p. 1226-1234 https://doi.org/10.1145/3219819.3219974
Chen, Yongjun ; Shi, Min ; Gao, Hongyang ; Shen, Dinggang ; Cai, Lei ; Ji, Shuiwang. / Voxel deconvolutional networks for 3D brain image labeling. KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Association for Computing Machinery, 2018. pp. 1226-1234
@inproceedings{8b6abb813641434ba17dd9dd8ca7d207,
title = "Voxel deconvolutional networks for 3D brain image labeling",
abstract = "Deep learning methods have shown great success in pixel-wise prediction tasks. One of the most popular methods employs an encoder-decoder network in which deconvolutional layers are used for up-sampling feature maps. However, a key limitation of the deconvolutional layer is that it suffers from the checkerboard artifact problem, which harms the prediction accuracy. This is caused by the independency among adjacent pixels on the output feature maps. Previous work only solved the checkerboard artifact issue of deconvolutional layers in the 2D space. Since the number of intermediate feature maps needed to generate a deconvolutional layer grows exponentially with dimensionality, it is more challenging to solve this issue in higher dimensions. In this work, we propose the voxel deconvolutional layer (VoxelDCL) to solve the checkerboard artifact problem of deconvolutional layers in 3D space. We also provide an efficient approach to implement VoxelDCL. To demonstrate the effectiveness of VoxelDCL, we build four variations of voxel deconvolutional networks (VoxelDCN) based on the U-Net architecture with VoxelDCL. We apply our networks to address volumetric brain images labeling tasks using the ADNI and Loni LPBA40 datasets. The experimental results show that the proposed iVoxelDCNa achieves improved performance in all experiments. It reaches 83.34{\%} in terms of dice ratio on the ADNI dataset and 79.12{\%} on the Loni LPBA40 dataset, which increases 1.39{\%} and 2.21{\%} respectively compared with the baseline. In addition, all the variations of VoxelDCN we proposed outperform the baseline methods on the above datasets, which demonstrates the effectiveness of our methods.",
keywords = "Deep learning, Volumetric brain image labeling, Voxel deconvolutional layer, Voxel deconvolutional networks",
author = "Yongjun Chen and Min Shi and Hongyang Gao and Dinggang Shen and Lei Cai and Shuiwang Ji",
year = "2018",
month = "7",
day = "19",
doi = "10.1145/3219819.3219974",
language = "English",
isbn = "9781450355520",
pages = "1226--1234",
booktitle = "KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining",
publisher = "Association for Computing Machinery",

}

TY - GEN

T1 - Voxel deconvolutional networks for 3D brain image labeling

AU - Chen, Yongjun

AU - Shi, Min

AU - Gao, Hongyang

AU - Shen, Dinggang

AU - Cai, Lei

AU - Ji, Shuiwang

PY - 2018/7/19

Y1 - 2018/7/19

N2 - Deep learning methods have shown great success in pixel-wise prediction tasks. One of the most popular methods employs an encoder-decoder network in which deconvolutional layers are used for up-sampling feature maps. However, a key limitation of the deconvolutional layer is that it suffers from the checkerboard artifact problem, which harms the prediction accuracy. This is caused by the independency among adjacent pixels on the output feature maps. Previous work only solved the checkerboard artifact issue of deconvolutional layers in the 2D space. Since the number of intermediate feature maps needed to generate a deconvolutional layer grows exponentially with dimensionality, it is more challenging to solve this issue in higher dimensions. In this work, we propose the voxel deconvolutional layer (VoxelDCL) to solve the checkerboard artifact problem of deconvolutional layers in 3D space. We also provide an efficient approach to implement VoxelDCL. To demonstrate the effectiveness of VoxelDCL, we build four variations of voxel deconvolutional networks (VoxelDCN) based on the U-Net architecture with VoxelDCL. We apply our networks to address volumetric brain images labeling tasks using the ADNI and Loni LPBA40 datasets. The experimental results show that the proposed iVoxelDCNa achieves improved performance in all experiments. It reaches 83.34% in terms of dice ratio on the ADNI dataset and 79.12% on the Loni LPBA40 dataset, which increases 1.39% and 2.21% respectively compared with the baseline. In addition, all the variations of VoxelDCN we proposed outperform the baseline methods on the above datasets, which demonstrates the effectiveness of our methods.

AB - Deep learning methods have shown great success in pixel-wise prediction tasks. One of the most popular methods employs an encoder-decoder network in which deconvolutional layers are used for up-sampling feature maps. However, a key limitation of the deconvolutional layer is that it suffers from the checkerboard artifact problem, which harms the prediction accuracy. This is caused by the independency among adjacent pixels on the output feature maps. Previous work only solved the checkerboard artifact issue of deconvolutional layers in the 2D space. Since the number of intermediate feature maps needed to generate a deconvolutional layer grows exponentially with dimensionality, it is more challenging to solve this issue in higher dimensions. In this work, we propose the voxel deconvolutional layer (VoxelDCL) to solve the checkerboard artifact problem of deconvolutional layers in 3D space. We also provide an efficient approach to implement VoxelDCL. To demonstrate the effectiveness of VoxelDCL, we build four variations of voxel deconvolutional networks (VoxelDCN) based on the U-Net architecture with VoxelDCL. We apply our networks to address volumetric brain images labeling tasks using the ADNI and Loni LPBA40 datasets. The experimental results show that the proposed iVoxelDCNa achieves improved performance in all experiments. It reaches 83.34% in terms of dice ratio on the ADNI dataset and 79.12% on the Loni LPBA40 dataset, which increases 1.39% and 2.21% respectively compared with the baseline. In addition, all the variations of VoxelDCN we proposed outperform the baseline methods on the above datasets, which demonstrates the effectiveness of our methods.

KW - Deep learning

KW - Volumetric brain image labeling

KW - Voxel deconvolutional layer

KW - Voxel deconvolutional networks

UR - http://www.scopus.com/inward/record.url?scp=85051533356&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85051533356&partnerID=8YFLogxK

U2 - 10.1145/3219819.3219974

DO - 10.1145/3219819.3219974

M3 - Conference contribution

AN - SCOPUS:85051533356

SN - 9781450355520

SP - 1226

EP - 1234

BT - KDD 2018 - Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining

PB - Association for Computing Machinery

ER -