Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data

Xiaofeng Zhu, Kim Han Thung, Ehsan Adeli, Yu Zhang, Dinggang Shen

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

4 Citations (Scopus)

Abstract

It is challenging to use incomplete multimodality data for Alzheimer’s Disease (AD) diagnosis. The current methods to address this challenge, such as low-rank matrix completion (i.e., imputing the missing values and unknown labels simultaneously) and multi-task learning (i.e., defining one regression task for each combination of modalities and then learning them jointly), are unable to model the complex data-to-label relationship in AD diagnosis and also ignore the heterogeneity among the modalities. In light of this, we propose a new Maximum Mean Discrepancy (MMD) based Multiple Kernel Learning (MKL) method for AD diagnosis using incomplete multimodality data. Specifically, we map all the samples from different modalities into a Reproducing Kernel Hilbert Space (RKHS), by devising a new MMD algorithm. The proposed MMD method incorporates data distribution matching, pair-wise sample matching and feature selection in an unified formulation, thus alleviating the modality heterogeneity issue and making all the samples comparable to share a common classifier in the RKHS. The resulting classifier obviously captures the nonlinear data-to-label relationship. We have tested our method using MRI and PET data from Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset for AD diagnosis. The experimental results show that our method outperforms other methods.

Original languageEnglish
Title of host publicationMedical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings
PublisherSpringer Verlag
Pages72-80
Number of pages9
Volume10435 LNCS
ISBN (Print)9783319661780
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)
Volume10435 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

Neuroimaging
Multimodality
Alzheimer's Disease
Discrepancy
kernel
Modality
Labels
Hilbert spaces
Reproducing Kernel Hilbert Space
Classifiers
Classifier
Multi-task Learning
Matrix Completion
Low-rank Matrices
Magnetic resonance imaging
Missing Values
Data Distribution
Feature extraction
Feature Selection
Learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Zhu, X., Thung, K. H., Adeli, E., Zhang, Y., & Shen, D. (2017). Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data. In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings (Vol. 10435 LNCS, pp. 72-80). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10435 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-66179-7_9

Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data. / Zhu, Xiaofeng; Thung, Kim Han; Adeli, Ehsan; Zhang, Yu; Shen, Dinggang.

Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS Springer Verlag, 2017. p. 72-80 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10435 LNCS).

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

Zhu, X, Thung, KH, Adeli, E, Zhang, Y & Shen, D 2017, Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data. in Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. vol. 10435 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10435 LNCS, Springer Verlag, pp. 72-80, 20th International Conference on Medical Image Computing and Computer-Assisted Intervention, MICCAI 2017, Quebec City, Canada, 17/9/11. https://doi.org/10.1007/978-3-319-66179-7_9
Zhu X, Thung KH, Adeli E, Zhang Y, Shen D. Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data. In Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS. Springer Verlag. 2017. p. 72-80. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-66179-7_9
Zhu, Xiaofeng ; Thung, Kim Han ; Adeli, Ehsan ; Zhang, Yu ; Shen, Dinggang. / Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data. Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings. Vol. 10435 LNCS Springer Verlag, 2017. pp. 72-80 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{944e3fd49f2847e697e193a695a2e4fe,
title = "Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data",
abstract = "It is challenging to use incomplete multimodality data for Alzheimer’s Disease (AD) diagnosis. The current methods to address this challenge, such as low-rank matrix completion (i.e., imputing the missing values and unknown labels simultaneously) and multi-task learning (i.e., defining one regression task for each combination of modalities and then learning them jointly), are unable to model the complex data-to-label relationship in AD diagnosis and also ignore the heterogeneity among the modalities. In light of this, we propose a new Maximum Mean Discrepancy (MMD) based Multiple Kernel Learning (MKL) method for AD diagnosis using incomplete multimodality data. Specifically, we map all the samples from different modalities into a Reproducing Kernel Hilbert Space (RKHS), by devising a new MMD algorithm. The proposed MMD method incorporates data distribution matching, pair-wise sample matching and feature selection in an unified formulation, thus alleviating the modality heterogeneity issue and making all the samples comparable to share a common classifier in the RKHS. The resulting classifier obviously captures the nonlinear data-to-label relationship. We have tested our method using MRI and PET data from Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset for AD diagnosis. The experimental results show that our method outperforms other methods.",
author = "Xiaofeng Zhu and Thung, {Kim Han} and Ehsan Adeli and Yu Zhang and Dinggang Shen",
year = "2017",
doi = "10.1007/978-3-319-66179-7_9",
language = "English",
isbn = "9783319661780",
volume = "10435 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "72--80",
booktitle = "Medical Image Computing and Computer Assisted Intervention − MICCAI 2017 - 20th International Conference, Proceedings",

}

TY - GEN

T1 - Maximum mean discrepancy based multiple kernel learning for incomplete multimodality neuroimaging data

AU - Zhu, Xiaofeng

AU - Thung, Kim Han

AU - Adeli, Ehsan

AU - Zhang, Yu

AU - Shen, Dinggang

PY - 2017

Y1 - 2017

N2 - It is challenging to use incomplete multimodality data for Alzheimer’s Disease (AD) diagnosis. The current methods to address this challenge, such as low-rank matrix completion (i.e., imputing the missing values and unknown labels simultaneously) and multi-task learning (i.e., defining one regression task for each combination of modalities and then learning them jointly), are unable to model the complex data-to-label relationship in AD diagnosis and also ignore the heterogeneity among the modalities. In light of this, we propose a new Maximum Mean Discrepancy (MMD) based Multiple Kernel Learning (MKL) method for AD diagnosis using incomplete multimodality data. Specifically, we map all the samples from different modalities into a Reproducing Kernel Hilbert Space (RKHS), by devising a new MMD algorithm. The proposed MMD method incorporates data distribution matching, pair-wise sample matching and feature selection in an unified formulation, thus alleviating the modality heterogeneity issue and making all the samples comparable to share a common classifier in the RKHS. The resulting classifier obviously captures the nonlinear data-to-label relationship. We have tested our method using MRI and PET data from Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset for AD diagnosis. The experimental results show that our method outperforms other methods.

AB - It is challenging to use incomplete multimodality data for Alzheimer’s Disease (AD) diagnosis. The current methods to address this challenge, such as low-rank matrix completion (i.e., imputing the missing values and unknown labels simultaneously) and multi-task learning (i.e., defining one regression task for each combination of modalities and then learning them jointly), are unable to model the complex data-to-label relationship in AD diagnosis and also ignore the heterogeneity among the modalities. In light of this, we propose a new Maximum Mean Discrepancy (MMD) based Multiple Kernel Learning (MKL) method for AD diagnosis using incomplete multimodality data. Specifically, we map all the samples from different modalities into a Reproducing Kernel Hilbert Space (RKHS), by devising a new MMD algorithm. The proposed MMD method incorporates data distribution matching, pair-wise sample matching and feature selection in an unified formulation, thus alleviating the modality heterogeneity issue and making all the samples comparable to share a common classifier in the RKHS. The resulting classifier obviously captures the nonlinear data-to-label relationship. We have tested our method using MRI and PET data from Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset for AD diagnosis. The experimental results show that our method outperforms other methods.

UR - http://www.scopus.com/inward/record.url?scp=85029509257&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85029509257&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-66179-7_9

DO - 10.1007/978-3-319-66179-7_9

M3 - Conference contribution

AN - SCOPUS:85029509257

SN - 9783319661780

VL - 10435 LNCS

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

SP - 72

EP - 80

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

PB - Springer Verlag

ER -