Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines

Oula Puonti, Koen Van Leemput, Jesper Duemose Nielsen, Christian Bauer, Hartwig Roman Siebner, Kristoffer Hougaard Madsen, Axel Thielscher

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

16 Downloads (Pure)

Abstract

Transcranial brain stimulation (TBS) techniques such as transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS) and others have seen a strong increase as tools in therapy and research within the last 20 years. In order to precisely target the stimulation, it is important to accurately model the individual head anatomy of a subject. Of particular importance is accurate reconstruction of the skull, as it has the strongest impact on the current pathways due to its low conductivity. Thus providing automated tools, which can reliably reconstruct the anatomy of the human head from magnetic resonance (MR) scans would be highly valuable for the application of transcranial stimulation methods. These head models can also be used to inform source localization methods such as EEG and MEG.Automated segmentation of the skull from MR images is, however, challenging as the skull emits very little signal in MR. In order to avoid topological defects, such as holes in the segmentations, a strong model of the skull shape is needed. In this paper we propose a new shape model for skull segmentation based on the so-called convolutional restricted Boltzmann machines (cRBMs). Compared to traditionally used lower-order shape models, such as pair-wise Markov random fields (MRFs), the cRBMs model local shapes in larger spatial neighborhoods while still allowing for efficient inference. We compare the skull segmentation accuracy of our approach to two previously published methods and show significant improvement.
Original languageEnglish
Title of host publicationMedical Imaging 2018: Image Processing
EditorsElsa D. Angelini , Bennett A. Landman
Number of pages8
Volume10574
PublisherSPIE - International Society for Optical Engineering
Publication date2018
DOIs
Publication statusPublished - 2018
EventSPIE Medical Imaging 2018 - Houston, United States
Duration: 10 Feb 201815 Feb 2018

Conference

ConferenceSPIE Medical Imaging 2018
CountryUnited States
CityHouston
Period10/02/201815/02/2018
SeriesProceedings of SPIE, the International Society for Optical Engineering
ISSN0277-786X

Keywords

  • Skull segmentation
  • MRI
  • Shape modeling
  • Transcranial brain stimulation
  • Head modeling

Cite this

Puonti, O., Van Leemput, K., Nielsen, J. D., Bauer, C., Siebner, H. R., Madsen, K. H., & Thielscher, A. (2018). Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines. In E. D. A., & B. A. L. (Eds.), Medical Imaging 2018: Image Processing (Vol. 10574). SPIE - International Society for Optical Engineering. Proceedings of SPIE, the International Society for Optical Engineering https://doi.org/10.1117/12.2293073
Puonti, Oula ; Van Leemput, Koen ; Nielsen, Jesper Duemose ; Bauer, Christian ; Siebner, Hartwig Roman ; Madsen, Kristoffer Hougaard ; Thielscher, Axel. / Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines. Medical Imaging 2018: Image Processing. editor / Elsa D. Angelini ; Bennett A. Landman. Vol. 10574 SPIE - International Society for Optical Engineering, 2018. (Proceedings of SPIE, the International Society for Optical Engineering).
@inproceedings{6a2f1199166848f0b4acabb4e3c2a6c3,
title = "Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines",
abstract = "Transcranial brain stimulation (TBS) techniques such as transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS) and others have seen a strong increase as tools in therapy and research within the last 20 years. In order to precisely target the stimulation, it is important to accurately model the individual head anatomy of a subject. Of particular importance is accurate reconstruction of the skull, as it has the strongest impact on the current pathways due to its low conductivity. Thus providing automated tools, which can reliably reconstruct the anatomy of the human head from magnetic resonance (MR) scans would be highly valuable for the application of transcranial stimulation methods. These head models can also be used to inform source localization methods such as EEG and MEG.Automated segmentation of the skull from MR images is, however, challenging as the skull emits very little signal in MR. In order to avoid topological defects, such as holes in the segmentations, a strong model of the skull shape is needed. In this paper we propose a new shape model for skull segmentation based on the so-called convolutional restricted Boltzmann machines (cRBMs). Compared to traditionally used lower-order shape models, such as pair-wise Markov random fields (MRFs), the cRBMs model local shapes in larger spatial neighborhoods while still allowing for efficient inference. We compare the skull segmentation accuracy of our approach to two previously published methods and show significant improvement.",
keywords = "Skull segmentation, MRI, Shape modeling, Transcranial brain stimulation, Head modeling",
author = "Oula Puonti and {Van Leemput}, Koen and Nielsen, {Jesper Duemose} and Christian Bauer and Siebner, {Hartwig Roman} and Madsen, {Kristoffer Hougaard} and Axel Thielscher",
year = "2018",
doi = "10.1117/12.2293073",
language = "English",
volume = "10574",
editor = "{Elsa D. Angelini} and { Bennett A. Landman}",
booktitle = "Medical Imaging 2018: Image Processing",
publisher = "SPIE - International Society for Optical Engineering",

}

Puonti, O, Van Leemput, K, Nielsen, JD, Bauer, C, Siebner, HR, Madsen, KH & Thielscher, A 2018, Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines. in EDA & BAL (eds), Medical Imaging 2018: Image Processing. vol. 10574, SPIE - International Society for Optical Engineering, Proceedings of SPIE, the International Society for Optical Engineering, SPIE Medical Imaging 2018, Houston, United States, 10/02/2018. https://doi.org/10.1117/12.2293073

Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines. / Puonti, Oula; Van Leemput, Koen; Nielsen, Jesper Duemose; Bauer, Christian; Siebner, Hartwig Roman; Madsen, Kristoffer Hougaard; Thielscher, Axel.

Medical Imaging 2018: Image Processing. ed. / Elsa D. Angelini; Bennett A. Landman. Vol. 10574 SPIE - International Society for Optical Engineering, 2018. (Proceedings of SPIE, the International Society for Optical Engineering).

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

TY - GEN

T1 - Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines

AU - Puonti, Oula

AU - Van Leemput, Koen

AU - Nielsen, Jesper Duemose

AU - Bauer, Christian

AU - Siebner, Hartwig Roman

AU - Madsen, Kristoffer Hougaard

AU - Thielscher, Axel

PY - 2018

Y1 - 2018

N2 - Transcranial brain stimulation (TBS) techniques such as transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS) and others have seen a strong increase as tools in therapy and research within the last 20 years. In order to precisely target the stimulation, it is important to accurately model the individual head anatomy of a subject. Of particular importance is accurate reconstruction of the skull, as it has the strongest impact on the current pathways due to its low conductivity. Thus providing automated tools, which can reliably reconstruct the anatomy of the human head from magnetic resonance (MR) scans would be highly valuable for the application of transcranial stimulation methods. These head models can also be used to inform source localization methods such as EEG and MEG.Automated segmentation of the skull from MR images is, however, challenging as the skull emits very little signal in MR. In order to avoid topological defects, such as holes in the segmentations, a strong model of the skull shape is needed. In this paper we propose a new shape model for skull segmentation based on the so-called convolutional restricted Boltzmann machines (cRBMs). Compared to traditionally used lower-order shape models, such as pair-wise Markov random fields (MRFs), the cRBMs model local shapes in larger spatial neighborhoods while still allowing for efficient inference. We compare the skull segmentation accuracy of our approach to two previously published methods and show significant improvement.

AB - Transcranial brain stimulation (TBS) techniques such as transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS) and others have seen a strong increase as tools in therapy and research within the last 20 years. In order to precisely target the stimulation, it is important to accurately model the individual head anatomy of a subject. Of particular importance is accurate reconstruction of the skull, as it has the strongest impact on the current pathways due to its low conductivity. Thus providing automated tools, which can reliably reconstruct the anatomy of the human head from magnetic resonance (MR) scans would be highly valuable for the application of transcranial stimulation methods. These head models can also be used to inform source localization methods such as EEG and MEG.Automated segmentation of the skull from MR images is, however, challenging as the skull emits very little signal in MR. In order to avoid topological defects, such as holes in the segmentations, a strong model of the skull shape is needed. In this paper we propose a new shape model for skull segmentation based on the so-called convolutional restricted Boltzmann machines (cRBMs). Compared to traditionally used lower-order shape models, such as pair-wise Markov random fields (MRFs), the cRBMs model local shapes in larger spatial neighborhoods while still allowing for efficient inference. We compare the skull segmentation accuracy of our approach to two previously published methods and show significant improvement.

KW - Skull segmentation

KW - MRI

KW - Shape modeling

KW - Transcranial brain stimulation

KW - Head modeling

U2 - 10.1117/12.2293073

DO - 10.1117/12.2293073

M3 - Article in proceedings

VL - 10574

BT - Medical Imaging 2018: Image Processing

A2 - , Elsa D. Angelini

A2 - , Bennett A. Landman

PB - SPIE - International Society for Optical Engineering

ER -

Puonti O, Van Leemput K, Nielsen JD, Bauer C, Siebner HR, Madsen KH et al. Skull segmentation from MR scans using a higher-order shape model based on convolutional restricted Boltzmann machines. In EDA, BAL, editors, Medical Imaging 2018: Image Processing. Vol. 10574. SPIE - International Society for Optical Engineering. 2018. (Proceedings of SPIE, the International Society for Optical Engineering). https://doi.org/10.1117/12.2293073