Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF)

Zhenghan Fang, Yong Chen, Mingxia Liu, Yiqiang Zhan, Weili Lin, Dinggang Shen

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

Abstract

Magnetic resonance fingerprinting (MRF) is a novel quantitative imaging technique that allows simultaneous measurements of multiple important tissue properties in human body, e.g., T1 and T2 relaxation times. While MRF has demonstrated better scan efficiency as compared to conventional quantitative imaging techniques, further acceleration is desired, especially for certain subjects such as infants and young children. However, the conventional MRF framework only uses a simple template matching algorithm to quantify tissue properties, without considering the underlying spatial association among pixels in MRF signals. In this work, we aim to accelerate MRF acquisition by developing a new post-processing method that allows accurate quantification of tissue properties with fewer sampling data. Moreover, to improve the accuracy in quantification, the MRF signals from multiple surrounding pixels are used together to better estimate tissue properties at the central target pixel, which was simply done with the signal only from the target pixel in the original template matching method. In particular, a deep learning model, i.e., U-Net, is used to learn the mapping from the MRF signal evolutions to the tissue property map. To further reduce the network size of U-Net, principal component analysis (PCA) is used to reduce the dimensionality of the input signals. Based on in vivo brain data, our method can achieve accurate quantification for both T1 and T2 by using only 25% time points, which are four times of acceleration in data acquisition compared to the original template matching method.

Original languageEnglish
Title of host publicationMachine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings
EditorsMingxia Liu, Heung-Il Suk, Yinghuan Shi
PublisherSpringer Verlag
Pages398-405
Number of pages8
ISBN (Print)9783030009182
DOIs
Publication statusPublished - 2018 Jan 1
Externally publishedYes
Event9th International Workshop on Machine Learning in Medical Imaging, MLMI 2018 held in conjunction with the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018 - Granada, Spain
Duration: 2018 Sep 162018 Sep 16

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11046 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other9th International Workshop on Machine Learning in Medical Imaging, MLMI 2018 held in conjunction with the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018
CountrySpain
CityGranada
Period18/9/1618/9/16

Fingerprint

Fingerprinting
Magnetic Resonance
Magnetic resonance
Quantification
Tissue
Template matching
Template Matching
Pixel
Pixels
Imaging
Imaging techniques
Target
Matching Algorithm
Learning
Deep learning
Data Acquisition
Post-processing
Relaxation Time
Relaxation time
Principal component analysis

Keywords

  • Deep learning
  • Magnetic resonance fingerprinting
  • Relaxation times

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Fang, Z., Chen, Y., Liu, M., Zhan, Y., Lin, W., & Shen, D. (2018). Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF). In M. Liu, H-I. Suk, & Y. Shi (Eds.), Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings (pp. 398-405). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11046 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-030-00919-9_46

Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF). / Fang, Zhenghan; Chen, Yong; Liu, Mingxia; Zhan, Yiqiang; Lin, Weili; Shen, Dinggang.

Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. ed. / Mingxia Liu; Heung-Il Suk; Yinghuan Shi. Springer Verlag, 2018. p. 398-405 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 11046 LNCS).

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

Fang, Z, Chen, Y, Liu, M, Zhan, Y, Lin, W & Shen, D 2018, Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF). in M Liu, H-I Suk & Y Shi (eds), Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 11046 LNCS, Springer Verlag, pp. 398-405, 9th International Workshop on Machine Learning in Medical Imaging, MLMI 2018 held in conjunction with the 21st International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2018, Granada, Spain, 18/9/16. https://doi.org/10.1007/978-3-030-00919-9_46
Fang Z, Chen Y, Liu M, Zhan Y, Lin W, Shen D. Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF). In Liu M, Suk H-I, Shi Y, editors, Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. Springer Verlag. 2018. p. 398-405. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-030-00919-9_46
Fang, Zhenghan ; Chen, Yong ; Liu, Mingxia ; Zhan, Yiqiang ; Lin, Weili ; Shen, Dinggang. / Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF). Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings. editor / Mingxia Liu ; Heung-Il Suk ; Yinghuan Shi. Springer Verlag, 2018. pp. 398-405 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{4e41d7d2ce894350a55318628a450f08,
title = "Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF)",
abstract = "Magnetic resonance fingerprinting (MRF) is a novel quantitative imaging technique that allows simultaneous measurements of multiple important tissue properties in human body, e.g., T1 and T2 relaxation times. While MRF has demonstrated better scan efficiency as compared to conventional quantitative imaging techniques, further acceleration is desired, especially for certain subjects such as infants and young children. However, the conventional MRF framework only uses a simple template matching algorithm to quantify tissue properties, without considering the underlying spatial association among pixels in MRF signals. In this work, we aim to accelerate MRF acquisition by developing a new post-processing method that allows accurate quantification of tissue properties with fewer sampling data. Moreover, to improve the accuracy in quantification, the MRF signals from multiple surrounding pixels are used together to better estimate tissue properties at the central target pixel, which was simply done with the signal only from the target pixel in the original template matching method. In particular, a deep learning model, i.e., U-Net, is used to learn the mapping from the MRF signal evolutions to the tissue property map. To further reduce the network size of U-Net, principal component analysis (PCA) is used to reduce the dimensionality of the input signals. Based on in vivo brain data, our method can achieve accurate quantification for both T1 and T2 by using only 25{\%} time points, which are four times of acceleration in data acquisition compared to the original template matching method.",
keywords = "Deep learning, Magnetic resonance fingerprinting, Relaxation times",
author = "Zhenghan Fang and Yong Chen and Mingxia Liu and Yiqiang Zhan and Weili Lin and Dinggang Shen",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-030-00919-9_46",
language = "English",
isbn = "9783030009182",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "398--405",
editor = "Mingxia Liu and Heung-Il Suk and Yinghuan Shi",
booktitle = "Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings",

}

TY - GEN

T1 - Deep learning for fast and spatially-constrained tissue quantification from highly-undersampled data in magnetic resonance fingerprinting (MRF)

AU - Fang, Zhenghan

AU - Chen, Yong

AU - Liu, Mingxia

AU - Zhan, Yiqiang

AU - Lin, Weili

AU - Shen, Dinggang

PY - 2018/1/1

Y1 - 2018/1/1

N2 - Magnetic resonance fingerprinting (MRF) is a novel quantitative imaging technique that allows simultaneous measurements of multiple important tissue properties in human body, e.g., T1 and T2 relaxation times. While MRF has demonstrated better scan efficiency as compared to conventional quantitative imaging techniques, further acceleration is desired, especially for certain subjects such as infants and young children. However, the conventional MRF framework only uses a simple template matching algorithm to quantify tissue properties, without considering the underlying spatial association among pixels in MRF signals. In this work, we aim to accelerate MRF acquisition by developing a new post-processing method that allows accurate quantification of tissue properties with fewer sampling data. Moreover, to improve the accuracy in quantification, the MRF signals from multiple surrounding pixels are used together to better estimate tissue properties at the central target pixel, which was simply done with the signal only from the target pixel in the original template matching method. In particular, a deep learning model, i.e., U-Net, is used to learn the mapping from the MRF signal evolutions to the tissue property map. To further reduce the network size of U-Net, principal component analysis (PCA) is used to reduce the dimensionality of the input signals. Based on in vivo brain data, our method can achieve accurate quantification for both T1 and T2 by using only 25% time points, which are four times of acceleration in data acquisition compared to the original template matching method.

AB - Magnetic resonance fingerprinting (MRF) is a novel quantitative imaging technique that allows simultaneous measurements of multiple important tissue properties in human body, e.g., T1 and T2 relaxation times. While MRF has demonstrated better scan efficiency as compared to conventional quantitative imaging techniques, further acceleration is desired, especially for certain subjects such as infants and young children. However, the conventional MRF framework only uses a simple template matching algorithm to quantify tissue properties, without considering the underlying spatial association among pixels in MRF signals. In this work, we aim to accelerate MRF acquisition by developing a new post-processing method that allows accurate quantification of tissue properties with fewer sampling data. Moreover, to improve the accuracy in quantification, the MRF signals from multiple surrounding pixels are used together to better estimate tissue properties at the central target pixel, which was simply done with the signal only from the target pixel in the original template matching method. In particular, a deep learning model, i.e., U-Net, is used to learn the mapping from the MRF signal evolutions to the tissue property map. To further reduce the network size of U-Net, principal component analysis (PCA) is used to reduce the dimensionality of the input signals. Based on in vivo brain data, our method can achieve accurate quantification for both T1 and T2 by using only 25% time points, which are four times of acceleration in data acquisition compared to the original template matching method.

KW - Deep learning

KW - Magnetic resonance fingerprinting

KW - Relaxation times

UR - http://www.scopus.com/inward/record.url?scp=85054543278&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054543278&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-00919-9_46

DO - 10.1007/978-3-030-00919-9_46

M3 - Conference contribution

SN - 9783030009182

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 398

EP - 405

BT - Machine Learning in Medical Imaging - 9th International Workshop, MLMI 2018, Held in Conjunction with MICCAI 2018, Proceedings

A2 - Liu, Mingxia

A2 - Suk, Heung-Il

A2 - Shi, Yinghuan

PB - Springer Verlag

ER -