Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks

Khosro Bahrami, Islem Rekik, Feng Shi, Dinggang Shen

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

5 Citations (Scopus)

Abstract

7T MRI scanner provides MR images with higher resolution and better contrast than 3T MR scanners. This helps many medical analysis tasks, including tissue segmentation. However, currently there is a very limited number of 7T MRI scanners worldwide. This motivates us to propose a novel image post-processing framework that can jointly generate high-resolution 7T-like images and their corresponding high-quality 7T-like tissue segmentation maps, solely from the routine 3T MR images. Our proposed framework comprises two parallel components, namely (1) reconstruction and (2) segmentation. The reconstruction component includes the multi-step cascaded convolutional neural networks (CNNs) that map the input 3T MR image to a 7T-like MR image, in terms of both resolution and contrast. Similarly, the segmentation component involves another paralleled cascaded CNNs, with a different architecture, to generate high-quality segmentation maps. These cascaded feedbacks between the two designed paralleled CNNs allow both tasks to mutually benefit from each another when learning the respective reconstruction and segmentation mappings. For evaluation, we have tested our framework on 15 subjects (with paired 3T and 7T images) using a leave-one-out cross-validation. The experimental results show that our estimated 7T-like images have richer anatomical details and better segmentation results, compared to the 3T MRI. Furthermore, our method also achieved better results in both reconstruction and segmentation tasks, compared to the state-of-the-art methods.

Original languageEnglish
Title of host publicationMedical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings
PublisherSpringer Verlag
Pages764-772
Number of pages9
Volume10433 LNCS
ISBN (Print)9783319661810
DOIs
Publication statusPublished - 2017
Externally publishedYes
Event20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017 - Quebec City, Canada
Duration: 2017 Sep 112017 Sep 13

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10433 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017
CountryCanada
CityQuebec City
Period17/9/1117/9/13

Fingerprint

Magnetic resonance imaging
Segmentation
Neural Networks
Neural networks
Tissue
Scanner
Feedback
High Resolution
Processing
Post-processing
Cross-validation
Evaluation
Experimental Results
Framework

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Bahrami, K., Rekik, I., Shi, F., & Shen, D. (2017). Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks. In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings (Vol. 10433 LNCS, pp. 764-772). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10433 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-66182-7_87

Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks. / Bahrami, Khosro; Rekik, Islem; Shi, Feng; Shen, Dinggang.

Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10433 LNCS Springer Verlag, 2017. p. 764-772 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10433 LNCS).

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

Bahrami, K, Rekik, I, Shi, F & Shen, D 2017, Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks. in Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. vol. 10433 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10433 LNCS, Springer Verlag, pp. 764-772, 20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017, Quebec City, Canada, 17/9/11. https://doi.org/10.1007/978-3-319-66182-7_87
Bahrami K, Rekik I, Shi F, Shen D. Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks. In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10433 LNCS. Springer Verlag. 2017. p. 764-772. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-66182-7_87
Bahrami, Khosro ; Rekik, Islem ; Shi, Feng ; Shen, Dinggang. / Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks. Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10433 LNCS Springer Verlag, 2017. pp. 764-772 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{21aa6a0580d749d8be869c07b7280244,
title = "Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks",
abstract = "7T MRI scanner provides MR images with higher resolution and better contrast than 3T MR scanners. This helps many medical analysis tasks, including tissue segmentation. However, currently there is a very limited number of 7T MRI scanners worldwide. This motivates us to propose a novel image post-processing framework that can jointly generate high-resolution 7T-like images and their corresponding high-quality 7T-like tissue segmentation maps, solely from the routine 3T MR images. Our proposed framework comprises two parallel components, namely (1) reconstruction and (2) segmentation. The reconstruction component includes the multi-step cascaded convolutional neural networks (CNNs) that map the input 3T MR image to a 7T-like MR image, in terms of both resolution and contrast. Similarly, the segmentation component involves another paralleled cascaded CNNs, with a different architecture, to generate high-quality segmentation maps. These cascaded feedbacks between the two designed paralleled CNNs allow both tasks to mutually benefit from each another when learning the respective reconstruction and segmentation mappings. For evaluation, we have tested our framework on 15 subjects (with paired 3T and 7T images) using a leave-one-out cross-validation. The experimental results show that our estimated 7T-like images have richer anatomical details and better segmentation results, compared to the 3T MRI. Furthermore, our method also achieved better results in both reconstruction and segmentation tasks, compared to the state-of-the-art methods.",
author = "Khosro Bahrami and Islem Rekik and Feng Shi and Dinggang Shen",
year = "2017",
doi = "10.1007/978-3-319-66182-7_87",
language = "English",
isbn = "9783319661810",
volume = "10433 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "764--772",
booktitle = "Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings",

}

TY - GEN

T1 - Joint reconstruction and segmentation of 7T-like MR images from 3T MRI based on cascaded convolutional neural networks

AU - Bahrami, Khosro

AU - Rekik, Islem

AU - Shi, Feng

AU - Shen, Dinggang

PY - 2017

Y1 - 2017

N2 - 7T MRI scanner provides MR images with higher resolution and better contrast than 3T MR scanners. This helps many medical analysis tasks, including tissue segmentation. However, currently there is a very limited number of 7T MRI scanners worldwide. This motivates us to propose a novel image post-processing framework that can jointly generate high-resolution 7T-like images and their corresponding high-quality 7T-like tissue segmentation maps, solely from the routine 3T MR images. Our proposed framework comprises two parallel components, namely (1) reconstruction and (2) segmentation. The reconstruction component includes the multi-step cascaded convolutional neural networks (CNNs) that map the input 3T MR image to a 7T-like MR image, in terms of both resolution and contrast. Similarly, the segmentation component involves another paralleled cascaded CNNs, with a different architecture, to generate high-quality segmentation maps. These cascaded feedbacks between the two designed paralleled CNNs allow both tasks to mutually benefit from each another when learning the respective reconstruction and segmentation mappings. For evaluation, we have tested our framework on 15 subjects (with paired 3T and 7T images) using a leave-one-out cross-validation. The experimental results show that our estimated 7T-like images have richer anatomical details and better segmentation results, compared to the 3T MRI. Furthermore, our method also achieved better results in both reconstruction and segmentation tasks, compared to the state-of-the-art methods.

AB - 7T MRI scanner provides MR images with higher resolution and better contrast than 3T MR scanners. This helps many medical analysis tasks, including tissue segmentation. However, currently there is a very limited number of 7T MRI scanners worldwide. This motivates us to propose a novel image post-processing framework that can jointly generate high-resolution 7T-like images and their corresponding high-quality 7T-like tissue segmentation maps, solely from the routine 3T MR images. Our proposed framework comprises two parallel components, namely (1) reconstruction and (2) segmentation. The reconstruction component includes the multi-step cascaded convolutional neural networks (CNNs) that map the input 3T MR image to a 7T-like MR image, in terms of both resolution and contrast. Similarly, the segmentation component involves another paralleled cascaded CNNs, with a different architecture, to generate high-quality segmentation maps. These cascaded feedbacks between the two designed paralleled CNNs allow both tasks to mutually benefit from each another when learning the respective reconstruction and segmentation mappings. For evaluation, we have tested our framework on 15 subjects (with paired 3T and 7T images) using a leave-one-out cross-validation. The experimental results show that our estimated 7T-like images have richer anatomical details and better segmentation results, compared to the 3T MRI. Furthermore, our method also achieved better results in both reconstruction and segmentation tasks, compared to the state-of-the-art methods.

UR - http://www.scopus.com/inward/record.url?scp=85029360284&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029360284&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-66182-7_87

DO - 10.1007/978-3-319-66182-7_87

M3 - Conference contribution

AN - SCOPUS:85029360284

SN - 9783319661810

VL - 10433 LNCS

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

SP - 764

EP - 772

BT - Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings

PB - Springer Verlag

ER -