Pediatric brain extraction from T2-weighted MR images using 3D dual frame U-net and human connectome database

Dongchan Kim, Jong Hee Chae, Yeji Han

Research output: Contribution to journalArticle

Abstract

Accurate extraction of brain tissues from magnetic resonance (MR) images is important in neuroradiology. However, brain extraction is more difficult for pediatric brains than for adult brains due to several factors including smaller brain sizes and lower tissue contrasts. In this work, we propose a brain extraction technique that utilizes dual frame (DF) 3D U-net deep learning architecture and the human connectome project (HCP) database for multislice 2D pediatric T2-weighted MR images with diseases. To improve segmentation accuracy in small pediatric brains with detailed boundary regions, DF 3D U-net architecture was used. We pretrained networks with the HCP database to compensate for the limited amount of MR images and manual segmentation masks of pediatric patients. For quantitative analysis, we compared the brain extraction results of brain extraction tool, DF, and conventional 3D U-net using the dice similarity coefficient (DSC), intersection of union (IoU), and boundary F1 (BF) scores; each deep learning architecture was evaluated with and without pretraining using the HCP. This study included 10 patients with diseases and all images were acquired using a PROPELLER MR sequence. Pretraining using the HCP database enhanced segmentation performance of the network, and the skip connections in the DF 3D U-net could enhance the contour similarity of segmentation results. Experimental results showed that the proposed method increased the DSC, IoU, and BF scores by 0.8%, 1.6%, and 1.5%, respectively, compared with those of the conventional 3D U-net without pretraining.

Original languageEnglish
Pages (from-to)476-482
Number of pages7
JournalInternational Journal of Imaging Systems and Technology
Volume29
Issue number4
DOIs
StatePublished - 1 Dec 2019

Fingerprint

Pediatrics
Magnetic resonance
Brain
Tissue
Masks

Keywords

  • HCP database
  • dual frame 3D U-net
  • pediatric brain extraction

Cite this

@article{4d279e1a248b4c7ea801d79919345bef,
title = "Pediatric brain extraction from T2-weighted MR images using 3D dual frame U-net and human connectome database",
abstract = "Accurate extraction of brain tissues from magnetic resonance (MR) images is important in neuroradiology. However, brain extraction is more difficult for pediatric brains than for adult brains due to several factors including smaller brain sizes and lower tissue contrasts. In this work, we propose a brain extraction technique that utilizes dual frame (DF) 3D U-net deep learning architecture and the human connectome project (HCP) database for multislice 2D pediatric T2-weighted MR images with diseases. To improve segmentation accuracy in small pediatric brains with detailed boundary regions, DF 3D U-net architecture was used. We pretrained networks with the HCP database to compensate for the limited amount of MR images and manual segmentation masks of pediatric patients. For quantitative analysis, we compared the brain extraction results of brain extraction tool, DF, and conventional 3D U-net using the dice similarity coefficient (DSC), intersection of union (IoU), and boundary F1 (BF) scores; each deep learning architecture was evaluated with and without pretraining using the HCP. This study included 10 patients with diseases and all images were acquired using a PROPELLER MR sequence. Pretraining using the HCP database enhanced segmentation performance of the network, and the skip connections in the DF 3D U-net could enhance the contour similarity of segmentation results. Experimental results showed that the proposed method increased the DSC, IoU, and BF scores by 0.8{\%}, 1.6{\%}, and 1.5{\%}, respectively, compared with those of the conventional 3D U-net without pretraining.",
keywords = "HCP database, dual frame 3D U-net, pediatric brain extraction",
author = "Dongchan Kim and Chae, {Jong Hee} and Yeji Han",
year = "2019",
month = "12",
day = "1",
doi = "10.1002/ima.22325",
language = "English",
volume = "29",
pages = "476--482",
journal = "International Journal of Imaging Systems and Technology",
issn = "0899-9457",
publisher = "John Wiley and Sons Inc.",
number = "4",

}

Pediatric brain extraction from T2-weighted MR images using 3D dual frame U-net and human connectome database. / Kim, Dongchan; Chae, Jong Hee; Han, Yeji.

In: International Journal of Imaging Systems and Technology, Vol. 29, No. 4, 01.12.2019, p. 476-482.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Pediatric brain extraction from T2-weighted MR images using 3D dual frame U-net and human connectome database

AU - Kim, Dongchan

AU - Chae, Jong Hee

AU - Han, Yeji

PY - 2019/12/1

Y1 - 2019/12/1

N2 - Accurate extraction of brain tissues from magnetic resonance (MR) images is important in neuroradiology. However, brain extraction is more difficult for pediatric brains than for adult brains due to several factors including smaller brain sizes and lower tissue contrasts. In this work, we propose a brain extraction technique that utilizes dual frame (DF) 3D U-net deep learning architecture and the human connectome project (HCP) database for multislice 2D pediatric T2-weighted MR images with diseases. To improve segmentation accuracy in small pediatric brains with detailed boundary regions, DF 3D U-net architecture was used. We pretrained networks with the HCP database to compensate for the limited amount of MR images and manual segmentation masks of pediatric patients. For quantitative analysis, we compared the brain extraction results of brain extraction tool, DF, and conventional 3D U-net using the dice similarity coefficient (DSC), intersection of union (IoU), and boundary F1 (BF) scores; each deep learning architecture was evaluated with and without pretraining using the HCP. This study included 10 patients with diseases and all images were acquired using a PROPELLER MR sequence. Pretraining using the HCP database enhanced segmentation performance of the network, and the skip connections in the DF 3D U-net could enhance the contour similarity of segmentation results. Experimental results showed that the proposed method increased the DSC, IoU, and BF scores by 0.8%, 1.6%, and 1.5%, respectively, compared with those of the conventional 3D U-net without pretraining.

AB - Accurate extraction of brain tissues from magnetic resonance (MR) images is important in neuroradiology. However, brain extraction is more difficult for pediatric brains than for adult brains due to several factors including smaller brain sizes and lower tissue contrasts. In this work, we propose a brain extraction technique that utilizes dual frame (DF) 3D U-net deep learning architecture and the human connectome project (HCP) database for multislice 2D pediatric T2-weighted MR images with diseases. To improve segmentation accuracy in small pediatric brains with detailed boundary regions, DF 3D U-net architecture was used. We pretrained networks with the HCP database to compensate for the limited amount of MR images and manual segmentation masks of pediatric patients. For quantitative analysis, we compared the brain extraction results of brain extraction tool, DF, and conventional 3D U-net using the dice similarity coefficient (DSC), intersection of union (IoU), and boundary F1 (BF) scores; each deep learning architecture was evaluated with and without pretraining using the HCP. This study included 10 patients with diseases and all images were acquired using a PROPELLER MR sequence. Pretraining using the HCP database enhanced segmentation performance of the network, and the skip connections in the DF 3D U-net could enhance the contour similarity of segmentation results. Experimental results showed that the proposed method increased the DSC, IoU, and BF scores by 0.8%, 1.6%, and 1.5%, respectively, compared with those of the conventional 3D U-net without pretraining.

KW - HCP database

KW - dual frame 3D U-net

KW - pediatric brain extraction

UR - http://www.scopus.com/inward/record.url?scp=85064179275&partnerID=8YFLogxK

U2 - 10.1002/ima.22325

DO - 10.1002/ima.22325

M3 - Article

AN - SCOPUS:85064179275

VL - 29

SP - 476

EP - 482

JO - International Journal of Imaging Systems and Technology

JF - International Journal of Imaging Systems and Technology

SN - 0899-9457

IS - 4

ER -