Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI

Shuyu Li, Feng Shi, Guangkai Ma, Minjeong Kim, Dinggang Shen

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

Abstract

Ultra-high field MR imaging (e.g., 7T) provides unprecedented spatial resolution and superior signal-to-noise ratio, compared to the routine MR imaging (e.g., 1.5T and 3T). It allows precise depiction of small anatomical structures such as hippocampus and further benefits diagnosis of neurodegenerative diseases. However, the routine MR imaging is still mainly used in research and clinical studies, where accurate hippocampus segmentation is desired. In this paper, we present an automatic method for segmenting hippocampus from the routine MR images by learning 7T-like features from the training 7T MR images. Our main idea is to map features of the routine MR image to be similar to 7T image features, thus increasing their discriminability in hippocampus segmentation. Specifically, we propose a patch-based mapping method to map image patches of the routine MR images to the space of image patches of the 7T MR images. Thus, for each patch in the routine MR image, we can generate a new mapped patch with 7T-like pattern. Then, using those mapped patches, we can use a random forest to train a sequence of classifiers for hippocampus segmentation based on the appearance, texture, and contexture features of those mapped patches. Finally, hippocampi in the test image can be segmented by applying the learned image patch mapping and trained classifiers. Experimental results show that the accuracy of hippocampus segmentation can be significantly improved by using our learned 7T-like image features, in comparison to the direct use of features extracted from the routine MR images.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages37-45
Number of pages9
Volume9467
ISBN (Print)9783319281933
DOIs
Publication statusPublished - 2015
Externally publishedYes
Event1st International Workshop on Patch-Based Techniques in Medical Imaging, Patch-MI 2015 - Munich, Germany
Duration: 2015 Oct 92015 Oct 9

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9467
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other1st International Workshop on Patch-Based Techniques in Medical Imaging, Patch-MI 2015
CountryGermany
CityMunich
Period15/10/915/10/9

Fingerprint

Hippocampus
Magnetic resonance imaging
Segmentation
Imaging techniques
Patch
Classifiers
Neurodegenerative diseases
Signal to noise ratio
Textures
Imaging
Classifier
Random Forest
Spatial Resolution
Texture

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Li, S., Shi, F., Ma, G., Kim, M., & Shen, D. (2015). Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9467, pp. 37-45). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9467). Springer Verlag. https://doi.org/10.1007/978-3-319-28194-0_5

Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI. / Li, Shuyu; Shi, Feng; Ma, Guangkai; Kim, Minjeong; Shen, Dinggang.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9467 Springer Verlag, 2015. p. 37-45 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9467).

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

Li, S, Shi, F, Ma, G, Kim, M & Shen, D 2015, Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 9467, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9467, Springer Verlag, pp. 37-45, 1st International Workshop on Patch-Based Techniques in Medical Imaging, Patch-MI 2015, Munich, Germany, 15/10/9. https://doi.org/10.1007/978-3-319-28194-0_5
Li S, Shi F, Ma G, Kim M, Shen D. Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9467. Springer Verlag. 2015. p. 37-45. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-28194-0_5
Li, Shuyu ; Shi, Feng ; Ma, Guangkai ; Kim, Minjeong ; Shen, Dinggang. / Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 9467 Springer Verlag, 2015. pp. 37-45 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{9597a38cd68d481c9014b5ad93540f13,
title = "Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI",
abstract = "Ultra-high field MR imaging (e.g., 7T) provides unprecedented spatial resolution and superior signal-to-noise ratio, compared to the routine MR imaging (e.g., 1.5T and 3T). It allows precise depiction of small anatomical structures such as hippocampus and further benefits diagnosis of neurodegenerative diseases. However, the routine MR imaging is still mainly used in research and clinical studies, where accurate hippocampus segmentation is desired. In this paper, we present an automatic method for segmenting hippocampus from the routine MR images by learning 7T-like features from the training 7T MR images. Our main idea is to map features of the routine MR image to be similar to 7T image features, thus increasing their discriminability in hippocampus segmentation. Specifically, we propose a patch-based mapping method to map image patches of the routine MR images to the space of image patches of the 7T MR images. Thus, for each patch in the routine MR image, we can generate a new mapped patch with 7T-like pattern. Then, using those mapped patches, we can use a random forest to train a sequence of classifiers for hippocampus segmentation based on the appearance, texture, and contexture features of those mapped patches. Finally, hippocampi in the test image can be segmented by applying the learned image patch mapping and trained classifiers. Experimental results show that the accuracy of hippocampus segmentation can be significantly improved by using our learned 7T-like image features, in comparison to the direct use of features extracted from the routine MR images.",
author = "Shuyu Li and Feng Shi and Guangkai Ma and Minjeong Kim and Dinggang Shen",
year = "2015",
doi = "10.1007/978-3-319-28194-0_5",
language = "English",
isbn = "9783319281933",
volume = "9467",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "37--45",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Improving accuracy of automatic hippocampus segmentation in routine MRI by features learned from ultra-high field MRI

AU - Li, Shuyu

AU - Shi, Feng

AU - Ma, Guangkai

AU - Kim, Minjeong

AU - Shen, Dinggang

PY - 2015

Y1 - 2015

N2 - Ultra-high field MR imaging (e.g., 7T) provides unprecedented spatial resolution and superior signal-to-noise ratio, compared to the routine MR imaging (e.g., 1.5T and 3T). It allows precise depiction of small anatomical structures such as hippocampus and further benefits diagnosis of neurodegenerative diseases. However, the routine MR imaging is still mainly used in research and clinical studies, where accurate hippocampus segmentation is desired. In this paper, we present an automatic method for segmenting hippocampus from the routine MR images by learning 7T-like features from the training 7T MR images. Our main idea is to map features of the routine MR image to be similar to 7T image features, thus increasing their discriminability in hippocampus segmentation. Specifically, we propose a patch-based mapping method to map image patches of the routine MR images to the space of image patches of the 7T MR images. Thus, for each patch in the routine MR image, we can generate a new mapped patch with 7T-like pattern. Then, using those mapped patches, we can use a random forest to train a sequence of classifiers for hippocampus segmentation based on the appearance, texture, and contexture features of those mapped patches. Finally, hippocampi in the test image can be segmented by applying the learned image patch mapping and trained classifiers. Experimental results show that the accuracy of hippocampus segmentation can be significantly improved by using our learned 7T-like image features, in comparison to the direct use of features extracted from the routine MR images.

AB - Ultra-high field MR imaging (e.g., 7T) provides unprecedented spatial resolution and superior signal-to-noise ratio, compared to the routine MR imaging (e.g., 1.5T and 3T). It allows precise depiction of small anatomical structures such as hippocampus and further benefits diagnosis of neurodegenerative diseases. However, the routine MR imaging is still mainly used in research and clinical studies, where accurate hippocampus segmentation is desired. In this paper, we present an automatic method for segmenting hippocampus from the routine MR images by learning 7T-like features from the training 7T MR images. Our main idea is to map features of the routine MR image to be similar to 7T image features, thus increasing their discriminability in hippocampus segmentation. Specifically, we propose a patch-based mapping method to map image patches of the routine MR images to the space of image patches of the 7T MR images. Thus, for each patch in the routine MR image, we can generate a new mapped patch with 7T-like pattern. Then, using those mapped patches, we can use a random forest to train a sequence of classifiers for hippocampus segmentation based on the appearance, texture, and contexture features of those mapped patches. Finally, hippocampi in the test image can be segmented by applying the learned image patch mapping and trained classifiers. Experimental results show that the accuracy of hippocampus segmentation can be significantly improved by using our learned 7T-like image features, in comparison to the direct use of features extracted from the routine MR images.

UR - http://www.scopus.com/inward/record.url?scp=84955255361&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84955255361&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-28194-0_5

DO - 10.1007/978-3-319-28194-0_5

M3 - Conference contribution

SN - 9783319281933

VL - 9467

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

SP - 37

EP - 45

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

PB - Springer Verlag

ER -