DeepHarmony: A deep learning approach to contrast harmonization across scanner changes

Blake E. Dewey, Can Zhao, Jacob C. Reinhold, A. Carass, Kathryn C. Fitzgerald, Elias S. Sotirchos, Shiv Saidha, J. Oh, Dzung L. Pham, Peter A. Calabresi, Peter C.M. van Zijl, Jerry L. Prince

Research output: Contribution to journalArticle

Abstract

Magnetic resonance imaging (MRI) is a flexible medical imaging modality that often lacks reproducibility between protocols and scanners. It has been shown that even when care is taken to standardize acquisitions, any changes in hardware, software, or protocol design can lead to differences in quantitative results. This greatly impacts the quantitative utility of MRI in multi-site or long-term studies, where consistency is often valued over image quality. We propose a method of contrast harmonization, called DeepHarmony, which uses a U-Net-based deep learning architecture to produce images with consistent contrast. To provide training data, a small overlap cohort (n = 8) was scanned using two different protocols. Images harmonized with DeepHarmony showed significant improvement in consistency of volume quantification between scanning protocols. A longitudinal MRI dataset of patients with multiple sclerosis was also used to evaluate the effect of a protocol change on atrophy calculations in a clinical research setting. The results show that atrophy calculations were substantially and significantly affected by protocol change, whereas such changes have a less significant effect and substantially reduced overall difference when using DeepHarmony. This establishes that DeepHarmony can be used with an overlap cohort to reduce inconsistencies in segmentation caused by changes in scanner protocol, allowing for modernization of hardware and protocol design in long-term studies without invalidating previously acquired data.

Original languageEnglish (US)
JournalMagnetic Resonance Imaging
DOIs
StatePublished - Jan 1 2019

Fingerprint

Magnetic resonance imaging
Magnetic Resonance Imaging
Learning
Atrophy
Hardware
Medical imaging
Diagnostic Imaging
Modernization
Social Change
Image quality
Multiple Sclerosis
Software
Scanning
Research
Deep learning
Datasets
caN protocol

Keywords

  • Contrast harmonization
  • Deep learning
  • Magnetic resonance imaging

ASJC Scopus subject areas

  • Biophysics
  • Biomedical Engineering
  • Radiology Nuclear Medicine and imaging

Cite this

DeepHarmony : A deep learning approach to contrast harmonization across scanner changes. / Dewey, Blake E.; Zhao, Can; Reinhold, Jacob C.; Carass, A.; Fitzgerald, Kathryn C.; Sotirchos, Elias S.; Saidha, Shiv; Oh, J.; Pham, Dzung L.; Calabresi, Peter A.; van Zijl, Peter C.M.; Prince, Jerry L.

In: Magnetic Resonance Imaging, 01.01.2019.

Research output: Contribution to journalArticle

@article{474970c2f71f4402a6bab1c85f46e1ee,
title = "DeepHarmony: A deep learning approach to contrast harmonization across scanner changes",
abstract = "Magnetic resonance imaging (MRI) is a flexible medical imaging modality that often lacks reproducibility between protocols and scanners. It has been shown that even when care is taken to standardize acquisitions, any changes in hardware, software, or protocol design can lead to differences in quantitative results. This greatly impacts the quantitative utility of MRI in multi-site or long-term studies, where consistency is often valued over image quality. We propose a method of contrast harmonization, called DeepHarmony, which uses a U-Net-based deep learning architecture to produce images with consistent contrast. To provide training data, a small overlap cohort (n = 8) was scanned using two different protocols. Images harmonized with DeepHarmony showed significant improvement in consistency of volume quantification between scanning protocols. A longitudinal MRI dataset of patients with multiple sclerosis was also used to evaluate the effect of a protocol change on atrophy calculations in a clinical research setting. The results show that atrophy calculations were substantially and significantly affected by protocol change, whereas such changes have a less significant effect and substantially reduced overall difference when using DeepHarmony. This establishes that DeepHarmony can be used with an overlap cohort to reduce inconsistencies in segmentation caused by changes in scanner protocol, allowing for modernization of hardware and protocol design in long-term studies without invalidating previously acquired data.",
keywords = "Contrast harmonization, Deep learning, Magnetic resonance imaging",
author = "Dewey, {Blake E.} and Can Zhao and Reinhold, {Jacob C.} and A. Carass and Fitzgerald, {Kathryn C.} and Sotirchos, {Elias S.} and Shiv Saidha and J. Oh and Pham, {Dzung L.} and Calabresi, {Peter A.} and {van Zijl}, {Peter C.M.} and Prince, {Jerry L.}",
year = "2019",
month = "1",
day = "1",
doi = "10.1016/j.mri.2019.05.041",
language = "English (US)",
journal = "Magnetic Resonance Imaging",
issn = "0730-725X",
publisher = "Elsevier Inc.",

}

TY - JOUR

T1 - DeepHarmony

T2 - A deep learning approach to contrast harmonization across scanner changes

AU - Dewey, Blake E.

AU - Zhao, Can

AU - Reinhold, Jacob C.

AU - Carass, A.

AU - Fitzgerald, Kathryn C.

AU - Sotirchos, Elias S.

AU - Saidha, Shiv

AU - Oh, J.

AU - Pham, Dzung L.

AU - Calabresi, Peter A.

AU - van Zijl, Peter C.M.

AU - Prince, Jerry L.

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Magnetic resonance imaging (MRI) is a flexible medical imaging modality that often lacks reproducibility between protocols and scanners. It has been shown that even when care is taken to standardize acquisitions, any changes in hardware, software, or protocol design can lead to differences in quantitative results. This greatly impacts the quantitative utility of MRI in multi-site or long-term studies, where consistency is often valued over image quality. We propose a method of contrast harmonization, called DeepHarmony, which uses a U-Net-based deep learning architecture to produce images with consistent contrast. To provide training data, a small overlap cohort (n = 8) was scanned using two different protocols. Images harmonized with DeepHarmony showed significant improvement in consistency of volume quantification between scanning protocols. A longitudinal MRI dataset of patients with multiple sclerosis was also used to evaluate the effect of a protocol change on atrophy calculations in a clinical research setting. The results show that atrophy calculations were substantially and significantly affected by protocol change, whereas such changes have a less significant effect and substantially reduced overall difference when using DeepHarmony. This establishes that DeepHarmony can be used with an overlap cohort to reduce inconsistencies in segmentation caused by changes in scanner protocol, allowing for modernization of hardware and protocol design in long-term studies without invalidating previously acquired data.

AB - Magnetic resonance imaging (MRI) is a flexible medical imaging modality that often lacks reproducibility between protocols and scanners. It has been shown that even when care is taken to standardize acquisitions, any changes in hardware, software, or protocol design can lead to differences in quantitative results. This greatly impacts the quantitative utility of MRI in multi-site or long-term studies, where consistency is often valued over image quality. We propose a method of contrast harmonization, called DeepHarmony, which uses a U-Net-based deep learning architecture to produce images with consistent contrast. To provide training data, a small overlap cohort (n = 8) was scanned using two different protocols. Images harmonized with DeepHarmony showed significant improvement in consistency of volume quantification between scanning protocols. A longitudinal MRI dataset of patients with multiple sclerosis was also used to evaluate the effect of a protocol change on atrophy calculations in a clinical research setting. The results show that atrophy calculations were substantially and significantly affected by protocol change, whereas such changes have a less significant effect and substantially reduced overall difference when using DeepHarmony. This establishes that DeepHarmony can be used with an overlap cohort to reduce inconsistencies in segmentation caused by changes in scanner protocol, allowing for modernization of hardware and protocol design in long-term studies without invalidating previously acquired data.

KW - Contrast harmonization

KW - Deep learning

KW - Magnetic resonance imaging

UR - http://www.scopus.com/inward/record.url?scp=85069945285&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069945285&partnerID=8YFLogxK

U2 - 10.1016/j.mri.2019.05.041

DO - 10.1016/j.mri.2019.05.041

M3 - Article

C2 - 31301354

AN - SCOPUS:85069945285

JO - Magnetic Resonance Imaging

JF - Magnetic Resonance Imaging

SN - 0730-725X

ER -