Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research

Daniel Tward, Anthony Kolasny, Fatima Khan, Juan C Troncoso, Michael Miller

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

Abstract

The Computational Anatomy Gateway, powered largely by the Comet (San Diego Supercomputer Center) and Stampede (Texas Advanced Computing Center) clusters through XSEDE, provides software as a service tools for atlas based analysis of human brain magnetic resonance images. This includes deformable registration, automatic labeling of tissue types, and morphometric analysis. Our goal is to extend these services to the broader neuroscience community, accommodating multiple model organisms and imaging modalities, as well as low quality or missing data. We developed a new approach to multimodality registration: by predicting one modality from another, we can replace ad hoc image similarity metrics (such as mutual information or normalized cross correlation) with a log likelihood under a noise model. This statistical approach enables us to account for missing data using the Expectation Maximization algorithm. For portability and scalability we have implemented this algorithm in tensorflow. For accessibility we have compiled and many working examples for multiple model organisms, imaging systems, and missing tissue or image anomaly situations. These examples are made easily usable in the form of Jupyter notebooks, and made publicly available through github. This framework will significantly reduce the barrier to entry for basic neuroscientists, enabling the community to benefit from atlas based computational image analysis techniques.

Original languageEnglish (US)
Title of host publicationProceedings of the Practice and Experience in Advanced Research Computing
Subtitle of host publicationRise of the Machines (Learning), PEARC 2019
PublisherAssociation for Computing Machinery
ISBN (Electronic)9781450372275
DOIs
StatePublished - Jul 28 2019
Event2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019 - Chicago, United States
Duration: Jul 28 2019Aug 1 2019

Publication series

NameACM International Conference Proceeding Series

Conference

Conference2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019
CountryUnited States
CityChicago
Period7/28/198/1/19

Fingerprint

Medical imaging
Tissue
Supercomputers
Magnetic resonance
Imaging systems
Labeling
Image analysis
Scalability
Brain
Imaging techniques

Keywords

  • Brain mapping
  • Image registration
  • Neuroimaging

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Computer Networks and Communications
  • Computer Vision and Pattern Recognition
  • Software

Cite this

Tward, D., Kolasny, A., Khan, F., Troncoso, J. C., & Miller, M. (2019). Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research. In Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019 [3332217] (ACM International Conference Proceeding Series). Association for Computing Machinery. https://doi.org/10.1145/3332186.3332217

Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research. / Tward, Daniel; Kolasny, Anthony; Khan, Fatima; Troncoso, Juan C; Miller, Michael.

Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019. Association for Computing Machinery, 2019. 3332217 (ACM International Conference Proceeding Series).

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

Tward, D, Kolasny, A, Khan, F, Troncoso, JC & Miller, M 2019, Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research. in Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019., 3332217, ACM International Conference Proceeding Series, Association for Computing Machinery, 2019 Conference on Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019, Chicago, United States, 7/28/19. https://doi.org/10.1145/3332186.3332217
Tward D, Kolasny A, Khan F, Troncoso JC, Miller M. Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research. In Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019. Association for Computing Machinery. 2019. 3332217. (ACM International Conference Proceeding Series). https://doi.org/10.1145/3332186.3332217
Tward, Daniel ; Kolasny, Anthony ; Khan, Fatima ; Troncoso, Juan C ; Miller, Michael. / Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research. Proceedings of the Practice and Experience in Advanced Research Computing: Rise of the Machines (Learning), PEARC 2019. Association for Computing Machinery, 2019. (ACM International Conference Proceeding Series).
@inproceedings{dffc20b4bed04e85bc883fa1470ef0ec,
title = "Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research",
abstract = "The Computational Anatomy Gateway, powered largely by the Comet (San Diego Supercomputer Center) and Stampede (Texas Advanced Computing Center) clusters through XSEDE, provides software as a service tools for atlas based analysis of human brain magnetic resonance images. This includes deformable registration, automatic labeling of tissue types, and morphometric analysis. Our goal is to extend these services to the broader neuroscience community, accommodating multiple model organisms and imaging modalities, as well as low quality or missing data. We developed a new approach to multimodality registration: by predicting one modality from another, we can replace ad hoc image similarity metrics (such as mutual information or normalized cross correlation) with a log likelihood under a noise model. This statistical approach enables us to account for missing data using the Expectation Maximization algorithm. For portability and scalability we have implemented this algorithm in tensorflow. For accessibility we have compiled and many working examples for multiple model organisms, imaging systems, and missing tissue or image anomaly situations. These examples are made easily usable in the form of Jupyter notebooks, and made publicly available through github. This framework will significantly reduce the barrier to entry for basic neuroscientists, enabling the community to benefit from atlas based computational image analysis techniques.",
keywords = "Brain mapping, Image registration, Neuroimaging",
author = "Daniel Tward and Anthony Kolasny and Fatima Khan and Troncoso, {Juan C} and Michael Miller",
year = "2019",
month = "7",
day = "28",
doi = "10.1145/3332186.3332217",
language = "English (US)",
series = "ACM International Conference Proceeding Series",
publisher = "Association for Computing Machinery",
booktitle = "Proceedings of the Practice and Experience in Advanced Research Computing",

}

TY - GEN

T1 - Expanding the Computational Anatomy Gateway from clinical imaging to basic neuroscience research

AU - Tward, Daniel

AU - Kolasny, Anthony

AU - Khan, Fatima

AU - Troncoso, Juan C

AU - Miller, Michael

PY - 2019/7/28

Y1 - 2019/7/28

N2 - The Computational Anatomy Gateway, powered largely by the Comet (San Diego Supercomputer Center) and Stampede (Texas Advanced Computing Center) clusters through XSEDE, provides software as a service tools for atlas based analysis of human brain magnetic resonance images. This includes deformable registration, automatic labeling of tissue types, and morphometric analysis. Our goal is to extend these services to the broader neuroscience community, accommodating multiple model organisms and imaging modalities, as well as low quality or missing data. We developed a new approach to multimodality registration: by predicting one modality from another, we can replace ad hoc image similarity metrics (such as mutual information or normalized cross correlation) with a log likelihood under a noise model. This statistical approach enables us to account for missing data using the Expectation Maximization algorithm. For portability and scalability we have implemented this algorithm in tensorflow. For accessibility we have compiled and many working examples for multiple model organisms, imaging systems, and missing tissue or image anomaly situations. These examples are made easily usable in the form of Jupyter notebooks, and made publicly available through github. This framework will significantly reduce the barrier to entry for basic neuroscientists, enabling the community to benefit from atlas based computational image analysis techniques.

AB - The Computational Anatomy Gateway, powered largely by the Comet (San Diego Supercomputer Center) and Stampede (Texas Advanced Computing Center) clusters through XSEDE, provides software as a service tools for atlas based analysis of human brain magnetic resonance images. This includes deformable registration, automatic labeling of tissue types, and morphometric analysis. Our goal is to extend these services to the broader neuroscience community, accommodating multiple model organisms and imaging modalities, as well as low quality or missing data. We developed a new approach to multimodality registration: by predicting one modality from another, we can replace ad hoc image similarity metrics (such as mutual information or normalized cross correlation) with a log likelihood under a noise model. This statistical approach enables us to account for missing data using the Expectation Maximization algorithm. For portability and scalability we have implemented this algorithm in tensorflow. For accessibility we have compiled and many working examples for multiple model organisms, imaging systems, and missing tissue or image anomaly situations. These examples are made easily usable in the form of Jupyter notebooks, and made publicly available through github. This framework will significantly reduce the barrier to entry for basic neuroscientists, enabling the community to benefit from atlas based computational image analysis techniques.

KW - Brain mapping

KW - Image registration

KW - Neuroimaging

UR - http://www.scopus.com/inward/record.url?scp=85071015265&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85071015265&partnerID=8YFLogxK

U2 - 10.1145/3332186.3332217

DO - 10.1145/3332186.3332217

M3 - Conference contribution

AN - SCOPUS:85071015265

T3 - ACM International Conference Proceeding Series

BT - Proceedings of the Practice and Experience in Advanced Research Computing

PB - Association for Computing Machinery

ER -