Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion

Kim Han Thung, Chong Yaw Wee, Pew Thian Yap, Dinggang Shen

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

6 Citations (Scopus)

Abstract

Incomplete dataset due to missing values is ubiquitous in multimodal neuroimaging data. Denoting an incomplete dataset as a feature matrix, where each row contains feature values of the multi-modality data of a sample, we propose a framework to predict the corresponding interrelated multiple target outputs (e.g., diagnosis label and clinical scores) from this feature matrix. This is achieved by applying a matrix completion algorithm on a shrunk version of the feature matrix that is augmented with the corresponding target output matrix, to simultaneously predict the missing feature values and the unknown target outputs. We shrink the matrix by first partition the large incomplete feature matrix into smaller submatrices that contain complete feature data. Treating each target output prediction from the submatrix as a task, we perform multi-task learning based feature and sample selections to select the most discriminative features and samples from each submatrix. Features and samples which are not selected from any of the submatrices are removed, resulting in a shrunk feature matrix, which is still incomplete. This shrunk matrix together with its corresponding target matrix (of possibly unknown values) are finally simultaneously completed using a low rank matrix completion algorithm. Experimental results using the ADNI dataset indicate that our proposed framework yields better identification accuracy at higher speed compared with conventional imputation-based identification methods.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages163-170
Number of pages8
Volume8184 LNCS
ISBN (Print)9783319022666
DOIs
Publication statusPublished - 2013 Jan 1
Externally publishedYes
Event4th International Workshop on Machine Learning in Medical Imaging, MLMI 2013, Held in Conjunction with 16th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2013 - Nagoya, Japan
Duration: 2013 Sep 222013 Sep 22

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8184 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other4th International Workshop on Machine Learning in Medical Imaging, MLMI 2013, Held in Conjunction with 16th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2013
CountryJapan
CityNagoya
Period13/9/2213/9/22

Fingerprint

Alzheimer's Disease
Shrinkage
Completion
Target
Matrix Completion
Output
Multi-task Learning
Low-rank Matrices
Sample Selection
Neuroimaging
Predict
Unknown
Multimodality
Missing Values
Imputation
Feature Selection
High Speed
Partition
Labels
Prediction

Keywords

  • ADNI
  • Alzheimer's disease
  • classification
  • incomplete data
  • Matrix completion

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Thung, K. H., Wee, C. Y., Yap, P. T., & Shen, D. (2013). Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8184 LNCS, pp. 163-170). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8184 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-02267-3_21

Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion. / Thung, Kim Han; Wee, Chong Yaw; Yap, Pew Thian; Shen, Dinggang.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8184 LNCS Springer Verlag, 2013. p. 163-170 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8184 LNCS).

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

Thung, KH, Wee, CY, Yap, PT & Shen, D 2013, Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 8184 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8184 LNCS, Springer Verlag, pp. 163-170, 4th International Workshop on Machine Learning in Medical Imaging, MLMI 2013, Held in Conjunction with 16th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2013, Nagoya, Japan, 13/9/22. https://doi.org/10.1007/978-3-319-02267-3_21
Thung KH, Wee CY, Yap PT, Shen D. Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8184 LNCS. Springer Verlag. 2013. p. 163-170. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-02267-3_21
Thung, Kim Han ; Wee, Chong Yaw ; Yap, Pew Thian ; Shen, Dinggang. / Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8184 LNCS Springer Verlag, 2013. pp. 163-170 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{f7a4aa3d5d3b46c38710e18bc050005e,
title = "Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion",
abstract = "Incomplete dataset due to missing values is ubiquitous in multimodal neuroimaging data. Denoting an incomplete dataset as a feature matrix, where each row contains feature values of the multi-modality data of a sample, we propose a framework to predict the corresponding interrelated multiple target outputs (e.g., diagnosis label and clinical scores) from this feature matrix. This is achieved by applying a matrix completion algorithm on a shrunk version of the feature matrix that is augmented with the corresponding target output matrix, to simultaneously predict the missing feature values and the unknown target outputs. We shrink the matrix by first partition the large incomplete feature matrix into smaller submatrices that contain complete feature data. Treating each target output prediction from the submatrix as a task, we perform multi-task learning based feature and sample selections to select the most discriminative features and samples from each submatrix. Features and samples which are not selected from any of the submatrices are removed, resulting in a shrunk feature matrix, which is still incomplete. This shrunk matrix together with its corresponding target matrix (of possibly unknown values) are finally simultaneously completed using a low rank matrix completion algorithm. Experimental results using the ADNI dataset indicate that our proposed framework yields better identification accuracy at higher speed compared with conventional imputation-based identification methods.",
keywords = "ADNI, Alzheimer's disease, classification, incomplete data, Matrix completion",
author = "Thung, {Kim Han} and Wee, {Chong Yaw} and Yap, {Pew Thian} and Dinggang Shen",
year = "2013",
month = "1",
day = "1",
doi = "10.1007/978-3-319-02267-3_21",
language = "English",
isbn = "9783319022666",
volume = "8184 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "163--170",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Identification of Alzheimer's disease using incomplete multimodal dataset via matrix shrinkage and completion

AU - Thung, Kim Han

AU - Wee, Chong Yaw

AU - Yap, Pew Thian

AU - Shen, Dinggang

PY - 2013/1/1

Y1 - 2013/1/1

N2 - Incomplete dataset due to missing values is ubiquitous in multimodal neuroimaging data. Denoting an incomplete dataset as a feature matrix, where each row contains feature values of the multi-modality data of a sample, we propose a framework to predict the corresponding interrelated multiple target outputs (e.g., diagnosis label and clinical scores) from this feature matrix. This is achieved by applying a matrix completion algorithm on a shrunk version of the feature matrix that is augmented with the corresponding target output matrix, to simultaneously predict the missing feature values and the unknown target outputs. We shrink the matrix by first partition the large incomplete feature matrix into smaller submatrices that contain complete feature data. Treating each target output prediction from the submatrix as a task, we perform multi-task learning based feature and sample selections to select the most discriminative features and samples from each submatrix. Features and samples which are not selected from any of the submatrices are removed, resulting in a shrunk feature matrix, which is still incomplete. This shrunk matrix together with its corresponding target matrix (of possibly unknown values) are finally simultaneously completed using a low rank matrix completion algorithm. Experimental results using the ADNI dataset indicate that our proposed framework yields better identification accuracy at higher speed compared with conventional imputation-based identification methods.

AB - Incomplete dataset due to missing values is ubiquitous in multimodal neuroimaging data. Denoting an incomplete dataset as a feature matrix, where each row contains feature values of the multi-modality data of a sample, we propose a framework to predict the corresponding interrelated multiple target outputs (e.g., diagnosis label and clinical scores) from this feature matrix. This is achieved by applying a matrix completion algorithm on a shrunk version of the feature matrix that is augmented with the corresponding target output matrix, to simultaneously predict the missing feature values and the unknown target outputs. We shrink the matrix by first partition the large incomplete feature matrix into smaller submatrices that contain complete feature data. Treating each target output prediction from the submatrix as a task, we perform multi-task learning based feature and sample selections to select the most discriminative features and samples from each submatrix. Features and samples which are not selected from any of the submatrices are removed, resulting in a shrunk feature matrix, which is still incomplete. This shrunk matrix together with its corresponding target matrix (of possibly unknown values) are finally simultaneously completed using a low rank matrix completion algorithm. Experimental results using the ADNI dataset indicate that our proposed framework yields better identification accuracy at higher speed compared with conventional imputation-based identification methods.

KW - ADNI

KW - Alzheimer's disease

KW - classification

KW - incomplete data

KW - Matrix completion

UR - http://www.scopus.com/inward/record.url?scp=84886743004&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84886743004&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-02267-3_21

DO - 10.1007/978-3-319-02267-3_21

M3 - Conference contribution

AN - SCOPUS:84886743004

SN - 9783319022666

VL - 8184 LNCS

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

SP - 163

EP - 170

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

PB - Springer Verlag

ER -