Parallel imaging in time-of-flight magnetic resonance angiography using deep multistream convolutional neural networks

Yohan Jun, Taejoon Eo, Hyungseob Shin, Taeseong Kim, Ho Joon Lee, Do Sik Hwang

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Purpose: To develop and evaluate a method of parallel imaging time-of-flight (TOF) MRA using deep multistream convolutional neural networks (CNNs). Methods: A deep parallel imaging network (“DPI-net”) was developed to reconstruct 3D multichannel MRA from undersampled data. It comprises 2 deep-learning networks: a network of multistream CNNs for extracting feature maps of multichannel images and a network of reconstruction CNNs for reconstructing images from the multistream network output feature maps. The images were evaluated using normalized root mean square error (NRMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) values, and the visibility of blood vessels was assessed by measuring the vessel sharpness of middle and posterior cerebral arteries on axial maximum intensity projection (MIP) images. Vessel sharpness was compared using paired t tests, between DPI-net, 2 conventional parallel imaging methods (SAKE and ESPIRiT), and a deep-learning method (U-net). Results: DPI-net showed superior performance in reconstructing vessel signals in both axial slices and MIP images for all reduction factors. This was supported by the quantitative metrics, with DPI-net showing the lowest NRMSE, the highest PSNR and SSIM (except R = 3.8 on sagittal MIP images, and R = 5.7 on axial slices and sagittal MIP images), and significantly higher vessel sharpness values than the other methods. Conclusion: DPI-net was effective in reconstructing 3D TOF MRA from highly undersampled multichannel MR data, achieving superior performance, both quantitatively and qualitatively, over conventional parallel imaging and other deep-learning methods.

Original languageEnglish
Pages (from-to)3840-3853
Number of pages14
JournalMagnetic Resonance in Medicine
Volume81
Issue number6
DOIs
Publication statusPublished - 2019 Jun 1

Fingerprint

Magnetic Resonance Angiography
Learning
Signal-To-Noise Ratio
Posterior Cerebral Artery
Middle Cerebral Artery
Blood Vessels

All Science Journal Classification (ASJC) codes

  • Radiology Nuclear Medicine and imaging

Cite this

Jun, Yohan ; Eo, Taejoon ; Shin, Hyungseob ; Kim, Taeseong ; Lee, Ho Joon ; Hwang, Do Sik. / Parallel imaging in time-of-flight magnetic resonance angiography using deep multistream convolutional neural networks. In: Magnetic Resonance in Medicine. 2019 ; Vol. 81, No. 6. pp. 3840-3853.
@article{8d110c638e1e41e9a4f4c7fa251672de,
title = "Parallel imaging in time-of-flight magnetic resonance angiography using deep multistream convolutional neural networks",
abstract = "Purpose: To develop and evaluate a method of parallel imaging time-of-flight (TOF) MRA using deep multistream convolutional neural networks (CNNs). Methods: A deep parallel imaging network (“DPI-net”) was developed to reconstruct 3D multichannel MRA from undersampled data. It comprises 2 deep-learning networks: a network of multistream CNNs for extracting feature maps of multichannel images and a network of reconstruction CNNs for reconstructing images from the multistream network output feature maps. The images were evaluated using normalized root mean square error (NRMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) values, and the visibility of blood vessels was assessed by measuring the vessel sharpness of middle and posterior cerebral arteries on axial maximum intensity projection (MIP) images. Vessel sharpness was compared using paired t tests, between DPI-net, 2 conventional parallel imaging methods (SAKE and ESPIRiT), and a deep-learning method (U-net). Results: DPI-net showed superior performance in reconstructing vessel signals in both axial slices and MIP images for all reduction factors. This was supported by the quantitative metrics, with DPI-net showing the lowest NRMSE, the highest PSNR and SSIM (except R = 3.8 on sagittal MIP images, and R = 5.7 on axial slices and sagittal MIP images), and significantly higher vessel sharpness values than the other methods. Conclusion: DPI-net was effective in reconstructing 3D TOF MRA from highly undersampled multichannel MR data, achieving superior performance, both quantitatively and qualitatively, over conventional parallel imaging and other deep-learning methods.",
author = "Yohan Jun and Taejoon Eo and Hyungseob Shin and Taeseong Kim and Lee, {Ho Joon} and Hwang, {Do Sik}",
year = "2019",
month = "6",
day = "1",
doi = "10.1002/mrm.27656",
language = "English",
volume = "81",
pages = "3840--3853",
journal = "Magnetic Resonance in Medicine",
issn = "0740-3194",
publisher = "John Wiley and Sons Inc.",
number = "6",

}

Parallel imaging in time-of-flight magnetic resonance angiography using deep multistream convolutional neural networks. / Jun, Yohan; Eo, Taejoon; Shin, Hyungseob; Kim, Taeseong; Lee, Ho Joon; Hwang, Do Sik.

In: Magnetic Resonance in Medicine, Vol. 81, No. 6, 01.06.2019, p. 3840-3853.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Parallel imaging in time-of-flight magnetic resonance angiography using deep multistream convolutional neural networks

AU - Jun, Yohan

AU - Eo, Taejoon

AU - Shin, Hyungseob

AU - Kim, Taeseong

AU - Lee, Ho Joon

AU - Hwang, Do Sik

PY - 2019/6/1

Y1 - 2019/6/1

N2 - Purpose: To develop and evaluate a method of parallel imaging time-of-flight (TOF) MRA using deep multistream convolutional neural networks (CNNs). Methods: A deep parallel imaging network (“DPI-net”) was developed to reconstruct 3D multichannel MRA from undersampled data. It comprises 2 deep-learning networks: a network of multistream CNNs for extracting feature maps of multichannel images and a network of reconstruction CNNs for reconstructing images from the multistream network output feature maps. The images were evaluated using normalized root mean square error (NRMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) values, and the visibility of blood vessels was assessed by measuring the vessel sharpness of middle and posterior cerebral arteries on axial maximum intensity projection (MIP) images. Vessel sharpness was compared using paired t tests, between DPI-net, 2 conventional parallel imaging methods (SAKE and ESPIRiT), and a deep-learning method (U-net). Results: DPI-net showed superior performance in reconstructing vessel signals in both axial slices and MIP images for all reduction factors. This was supported by the quantitative metrics, with DPI-net showing the lowest NRMSE, the highest PSNR and SSIM (except R = 3.8 on sagittal MIP images, and R = 5.7 on axial slices and sagittal MIP images), and significantly higher vessel sharpness values than the other methods. Conclusion: DPI-net was effective in reconstructing 3D TOF MRA from highly undersampled multichannel MR data, achieving superior performance, both quantitatively and qualitatively, over conventional parallel imaging and other deep-learning methods.

AB - Purpose: To develop and evaluate a method of parallel imaging time-of-flight (TOF) MRA using deep multistream convolutional neural networks (CNNs). Methods: A deep parallel imaging network (“DPI-net”) was developed to reconstruct 3D multichannel MRA from undersampled data. It comprises 2 deep-learning networks: a network of multistream CNNs for extracting feature maps of multichannel images and a network of reconstruction CNNs for reconstructing images from the multistream network output feature maps. The images were evaluated using normalized root mean square error (NRMSE), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM) values, and the visibility of blood vessels was assessed by measuring the vessel sharpness of middle and posterior cerebral arteries on axial maximum intensity projection (MIP) images. Vessel sharpness was compared using paired t tests, between DPI-net, 2 conventional parallel imaging methods (SAKE and ESPIRiT), and a deep-learning method (U-net). Results: DPI-net showed superior performance in reconstructing vessel signals in both axial slices and MIP images for all reduction factors. This was supported by the quantitative metrics, with DPI-net showing the lowest NRMSE, the highest PSNR and SSIM (except R = 3.8 on sagittal MIP images, and R = 5.7 on axial slices and sagittal MIP images), and significantly higher vessel sharpness values than the other methods. Conclusion: DPI-net was effective in reconstructing 3D TOF MRA from highly undersampled multichannel MR data, achieving superior performance, both quantitatively and qualitatively, over conventional parallel imaging and other deep-learning methods.

UR - http://www.scopus.com/inward/record.url?scp=85060332050&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85060332050&partnerID=8YFLogxK

U2 - 10.1002/mrm.27656

DO - 10.1002/mrm.27656

M3 - Article

C2 - 30666723

AN - SCOPUS:85060332050

VL - 81

SP - 3840

EP - 3853

JO - Magnetic Resonance in Medicine

JF - Magnetic Resonance in Medicine

SN - 0740-3194

IS - 6

ER -