Enhancement of Perivascular Spaces Using Densely Connected Deep Convolutional Neural Network

Euijin Jung, Philip Chikontwe, Xiaopeng Zong, Weili Lin, Dinggang Shen, Sang Hyun Park

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Perivascular spaces (PVS) in the human brain are related to various brain diseases. However, it is difficult to quantify them due to their thin and blurry appearance. In this paper, we introduce a deep-learning-based method, which can enhance a magnetic resonance (MR) image to better visualize the PVS. To accurately predict the enhanced image, we propose a very deep 3D convolutional neural network that contains densely connected networks with skip connections. The proposed networks can utilize rich contextual information derived from low-level to high-level features and effectively alleviate the gradient vanishing problem caused by the deep layers. The proposed method is evaluated on 17 7T MR images by a twofold cross-validation. The experiments show that our proposed network is much more effective to enhance the PVS than the previous PVS enhancement methods.

Original languageEnglish
Article number8632900
Pages (from-to)18382-18391
Number of pages10
JournalIEEE Access
Volume7
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

Magnetic resonance
Brain
Neural networks
Experiments
Deep learning

Keywords

  • deep convolutional neural network
  • densely connected network
  • MRI enhancement
  • Perivascular spaces
  • skip connections

ASJC Scopus subject areas

  • Computer Science(all)
  • Materials Science(all)
  • Engineering(all)

Cite this

Enhancement of Perivascular Spaces Using Densely Connected Deep Convolutional Neural Network. / Jung, Euijin; Chikontwe, Philip; Zong, Xiaopeng; Lin, Weili; Shen, Dinggang; Park, Sang Hyun.

In: IEEE Access, Vol. 7, 8632900, 01.01.2019, p. 18382-18391.

Research output: Contribution to journalArticle

Jung, Euijin ; Chikontwe, Philip ; Zong, Xiaopeng ; Lin, Weili ; Shen, Dinggang ; Park, Sang Hyun. / Enhancement of Perivascular Spaces Using Densely Connected Deep Convolutional Neural Network. In: IEEE Access. 2019 ; Vol. 7. pp. 18382-18391.
@article{c1562145f62942eab4f129acdcd5b4bd,
title = "Enhancement of Perivascular Spaces Using Densely Connected Deep Convolutional Neural Network",
abstract = "Perivascular spaces (PVS) in the human brain are related to various brain diseases. However, it is difficult to quantify them due to their thin and blurry appearance. In this paper, we introduce a deep-learning-based method, which can enhance a magnetic resonance (MR) image to better visualize the PVS. To accurately predict the enhanced image, we propose a very deep 3D convolutional neural network that contains densely connected networks with skip connections. The proposed networks can utilize rich contextual information derived from low-level to high-level features and effectively alleviate the gradient vanishing problem caused by the deep layers. The proposed method is evaluated on 17 7T MR images by a twofold cross-validation. The experiments show that our proposed network is much more effective to enhance the PVS than the previous PVS enhancement methods.",
keywords = "deep convolutional neural network, densely connected network, MRI enhancement, Perivascular spaces, skip connections",
author = "Euijin Jung and Philip Chikontwe and Xiaopeng Zong and Weili Lin and Dinggang Shen and Park, {Sang Hyun}",
year = "2019",
month = "1",
day = "1",
doi = "10.1109/ACCESS.2019.2896911",
language = "English",
volume = "7",
pages = "18382--18391",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - JOUR

T1 - Enhancement of Perivascular Spaces Using Densely Connected Deep Convolutional Neural Network

AU - Jung, Euijin

AU - Chikontwe, Philip

AU - Zong, Xiaopeng

AU - Lin, Weili

AU - Shen, Dinggang

AU - Park, Sang Hyun

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Perivascular spaces (PVS) in the human brain are related to various brain diseases. However, it is difficult to quantify them due to their thin and blurry appearance. In this paper, we introduce a deep-learning-based method, which can enhance a magnetic resonance (MR) image to better visualize the PVS. To accurately predict the enhanced image, we propose a very deep 3D convolutional neural network that contains densely connected networks with skip connections. The proposed networks can utilize rich contextual information derived from low-level to high-level features and effectively alleviate the gradient vanishing problem caused by the deep layers. The proposed method is evaluated on 17 7T MR images by a twofold cross-validation. The experiments show that our proposed network is much more effective to enhance the PVS than the previous PVS enhancement methods.

AB - Perivascular spaces (PVS) in the human brain are related to various brain diseases. However, it is difficult to quantify them due to their thin and blurry appearance. In this paper, we introduce a deep-learning-based method, which can enhance a magnetic resonance (MR) image to better visualize the PVS. To accurately predict the enhanced image, we propose a very deep 3D convolutional neural network that contains densely connected networks with skip connections. The proposed networks can utilize rich contextual information derived from low-level to high-level features and effectively alleviate the gradient vanishing problem caused by the deep layers. The proposed method is evaluated on 17 7T MR images by a twofold cross-validation. The experiments show that our proposed network is much more effective to enhance the PVS than the previous PVS enhancement methods.

KW - deep convolutional neural network

KW - densely connected network

KW - MRI enhancement

KW - Perivascular spaces

KW - skip connections

UR - http://www.scopus.com/inward/record.url?scp=85062226110&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85062226110&partnerID=8YFLogxK

U2 - 10.1109/ACCESS.2019.2896911

DO - 10.1109/ACCESS.2019.2896911

M3 - Article

AN - SCOPUS:85062226110

VL - 7

SP - 18382

EP - 18391

JO - IEEE Access

JF - IEEE Access

SN - 2169-3536

M1 - 8632900

ER -