Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy

Erdem Varol, Christos Davatzikos

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

Abstract

An important prerequisite for computational neuroanatomy is the spatial normalization of the data. Despite its importance for the success of the subsequent statistical analysis, image alignment is dealt with from the perspective of image matching, while its influence on the group analysis is neglected. The choice of the template, the registration algorithm as well as the registration parameters, all confound group differences and impact the outcome of the analysis. In order to limit their influence, we perform multiple registrations by varying these parameters, resulting in multiple instances for each sample. In order to harness the high dimensionality of the data and emphasize the group differences, we propose a supervised dimensionality reduction technique that takes into account the organization of the data. This is achieved by solving a supervised dictionary learning problem for block-sparse signals. Structured sparsity allows the grouping of instances across different independent samples, while label supervision allows for discriminative dictionaries. The block structure of dictionaries allows constructing multiple classifiers that treat each dictionary block as a basis of a subspace that spans a separate band of information. We formulate this problem as a convex optimization problem with a geometric programming (GP) component. Promising results that demonstrate the potential of the proposed approach are shown for an MR image dataset of Autism subjects.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages446-453
Number of pages8
Volume8674 LNCS
EditionPART 2
ISBN (Print)9783319104690
DOIs
StatePublished - 2014
Externally publishedYes
Event17th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2014 - Boston, MA, United States
Duration: Sep 14 2014Sep 18 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume8674 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other17th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2014
CountryUnited States
CityBoston, MA
Period9/14/149/18/14

Fingerprint

Anatomy
Glossaries
Clustering
Registration
Multiple Classifiers
Geometric Programming
Block Structure
Image matching
Image Matching
Convex optimization
Dimensionality Reduction
Convex Optimization
Sparsity
Grouping
Normalization
Statistical Analysis
Dimensionality
Labels
Template
Statistical methods

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Varol, E., & Davatzikos, C. (2014). Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 2 ed., Vol. 8674 LNCS, pp. 446-453). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8674 LNCS, No. PART 2). Springer Verlag. https://doi.org/10.1007/978-3-319-10470-6_56

Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy. / Varol, Erdem; Davatzikos, Christos.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8674 LNCS PART 2. ed. Springer Verlag, 2014. p. 446-453 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8674 LNCS, No. PART 2).

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

Varol, E & Davatzikos, C 2014, Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 2 edn, vol. 8674 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 8674 LNCS, Springer Verlag, pp. 446-453, 17th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2014, Boston, MA, United States, 9/14/14. https://doi.org/10.1007/978-3-319-10470-6_56
Varol E, Davatzikos C. Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 2 ed. Vol. 8674 LNCS. Springer Verlag. 2014. p. 446-453. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2). https://doi.org/10.1007/978-3-319-10470-6_56
Varol, Erdem ; Davatzikos, Christos. / Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8674 LNCS PART 2. ed. Springer Verlag, 2014. pp. 446-453 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
@inproceedings{4abb59fd506c470eaeaf499d56e6508e,
title = "Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy",
abstract = "An important prerequisite for computational neuroanatomy is the spatial normalization of the data. Despite its importance for the success of the subsequent statistical analysis, image alignment is dealt with from the perspective of image matching, while its influence on the group analysis is neglected. The choice of the template, the registration algorithm as well as the registration parameters, all confound group differences and impact the outcome of the analysis. In order to limit their influence, we perform multiple registrations by varying these parameters, resulting in multiple instances for each sample. In order to harness the high dimensionality of the data and emphasize the group differences, we propose a supervised dimensionality reduction technique that takes into account the organization of the data. This is achieved by solving a supervised dictionary learning problem for block-sparse signals. Structured sparsity allows the grouping of instances across different independent samples, while label supervision allows for discriminative dictionaries. The block structure of dictionaries allows constructing multiple classifiers that treat each dictionary block as a basis of a subspace that spans a separate band of information. We formulate this problem as a convex optimization problem with a geometric programming (GP) component. Promising results that demonstrate the potential of the proposed approach are shown for an MR image dataset of Autism subjects.",
author = "Erdem Varol and Christos Davatzikos",
year = "2014",
doi = "10.1007/978-3-319-10470-6_56",
language = "English (US)",
isbn = "9783319104690",
volume = "8674 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
number = "PART 2",
pages = "446--453",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 2",

}

TY - GEN

T1 - Supervised block sparse dictionary learning for simultaneous clustering and classification in computational anatomy

AU - Varol, Erdem

AU - Davatzikos, Christos

PY - 2014

Y1 - 2014

N2 - An important prerequisite for computational neuroanatomy is the spatial normalization of the data. Despite its importance for the success of the subsequent statistical analysis, image alignment is dealt with from the perspective of image matching, while its influence on the group analysis is neglected. The choice of the template, the registration algorithm as well as the registration parameters, all confound group differences and impact the outcome of the analysis. In order to limit their influence, we perform multiple registrations by varying these parameters, resulting in multiple instances for each sample. In order to harness the high dimensionality of the data and emphasize the group differences, we propose a supervised dimensionality reduction technique that takes into account the organization of the data. This is achieved by solving a supervised dictionary learning problem for block-sparse signals. Structured sparsity allows the grouping of instances across different independent samples, while label supervision allows for discriminative dictionaries. The block structure of dictionaries allows constructing multiple classifiers that treat each dictionary block as a basis of a subspace that spans a separate band of information. We formulate this problem as a convex optimization problem with a geometric programming (GP) component. Promising results that demonstrate the potential of the proposed approach are shown for an MR image dataset of Autism subjects.

AB - An important prerequisite for computational neuroanatomy is the spatial normalization of the data. Despite its importance for the success of the subsequent statistical analysis, image alignment is dealt with from the perspective of image matching, while its influence on the group analysis is neglected. The choice of the template, the registration algorithm as well as the registration parameters, all confound group differences and impact the outcome of the analysis. In order to limit their influence, we perform multiple registrations by varying these parameters, resulting in multiple instances for each sample. In order to harness the high dimensionality of the data and emphasize the group differences, we propose a supervised dimensionality reduction technique that takes into account the organization of the data. This is achieved by solving a supervised dictionary learning problem for block-sparse signals. Structured sparsity allows the grouping of instances across different independent samples, while label supervision allows for discriminative dictionaries. The block structure of dictionaries allows constructing multiple classifiers that treat each dictionary block as a basis of a subspace that spans a separate band of information. We formulate this problem as a convex optimization problem with a geometric programming (GP) component. Promising results that demonstrate the potential of the proposed approach are shown for an MR image dataset of Autism subjects.

UR - http://www.scopus.com/inward/record.url?scp=84906968592&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84906968592&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-10470-6_56

DO - 10.1007/978-3-319-10470-6_56

M3 - Conference contribution

SN - 9783319104690

VL - 8674 LNCS

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

SP - 446

EP - 453

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

PB - Springer Verlag

ER -