Cooperative Correlational and Discriminative Ensemble Classifier Learning for Early Dementia Diagnosis Using Morphological Brain Multiplexes

Rory Raeper, Anna Lisowska, Islem Rekik (Lead / Corresponding author)

Research output: Contribution to journalArticle

Abstract

Dementia alters the brain wiring on different levels. However, these changes might be subtle particularly in patients with early mild cognitive impairment (eMCI). Hence, developing accurate diagnostic techniques for eMCI identification is critical for early intervention to prevent the onset of Alzheimer’s Disease (AD). There is a large body of machine-learning based research developed for classifying different brain states (e.g., AD vs MCI) using neuroimaging data. These works can be fundamentally grouped into two categories. The first one uses correlational methods, such as canonical correlation analysis (CCA) and its variants, with the aim to identify most correlated features for diagnosis. The second one includes discriminative methods, such as feature selection methods and linear discriminative analysis (LDA) and its variants to identify brain features that discriminate between two brain states. However, existing methods examine these correlational and discriminative brain data independently, which overlooks the complementary information provided by both techniques, which could prove to be useful in data classification tasks. On the other hand, how early dementia affects cortical brain connections in morphology remains largely unexplored. To address these limitations, we propose a cooperative correlational and discriminative ensemble learning framework for eMCI diagnosis that leverages a brain network representation from multiple morphological networks, each derived from the cortical surface. Specifically, we devise ‘the shallow convolutional brain multiplex’ (SCBM), which encodes both region-to-region and network-to-network relationships. Then, we represent each individual brain using a set of SCBMs, which are used to train an ensemble of CCA-SVM and LDA-based classifiers, cooperating to output the label for a new testing subject. Overall, our framework outperformed several state-of-the-art methods including independent correlational and discriminative methods.
LanguageEnglish
Pages43830-43839
Number of pages10
JournalIEEE Access
Volume6
Early online date6 Aug 2018
DOIs
Publication statusPublished - 28 Aug 2018

Fingerprint

Brain
Classifiers
Neuroimaging
Electric wiring
Learning systems
Feature extraction
Labels
Testing

Keywords

  • Morphological Brian Network
  • Multi-View Brain Data
  • Canonical Correlation Analysis
  • Discriminative Methds
  • Linear Discriminant Analysis
  • Ensemble Classifier
  • Brain Multiplex

Cite this

@article{3852ee92ca8c4bb7b50b87751445deba,
title = "Cooperative Correlational and Discriminative Ensemble Classifier Learning for Early Dementia Diagnosis Using Morphological Brain Multiplexes",
abstract = "Dementia alters the brain wiring on different levels. However, these changes might be subtle particularly in patients with early mild cognitive impairment (eMCI). Hence, developing accurate diagnostic techniques for eMCI identification is critical for early intervention to prevent the onset of Alzheimer’s Disease (AD). There is a large body of machine-learning based research developed for classifying different brain states (e.g., AD vs MCI) using neuroimaging data. These works can be fundamentally grouped into two categories. The first one uses correlational methods, such as canonical correlation analysis (CCA) and its variants, with the aim to identify most correlated features for diagnosis. The second one includes discriminative methods, such as feature selection methods and linear discriminative analysis (LDA) and its variants to identify brain features that discriminate between two brain states. However, existing methods examine these correlational and discriminative brain data independently, which overlooks the complementary information provided by both techniques, which could prove to be useful in data classification tasks. On the other hand, how early dementia affects cortical brain connections in morphology remains largely unexplored. To address these limitations, we propose a cooperative correlational and discriminative ensemble learning framework for eMCI diagnosis that leverages a brain network representation from multiple morphological networks, each derived from the cortical surface. Specifically, we devise ‘the shallow convolutional brain multiplex’ (SCBM), which encodes both region-to-region and network-to-network relationships. Then, we represent each individual brain using a set of SCBMs, which are used to train an ensemble of CCA-SVM and LDA-based classifiers, cooperating to output the label for a new testing subject. Overall, our framework outperformed several state-of-the-art methods including independent correlational and discriminative methods.",
keywords = "Morphological Brian Network, Multi-View Brain Data, Canonical Correlation Analysis, Discriminative Methds, Linear Discriminant Analysis, Ensemble Classifier, Brain Multiplex",
author = "Rory Raeper and Anna Lisowska and Islem Rekik",
year = "2018",
month = "8",
day = "28",
doi = "10.1109/ACCESS.2018.2863657",
language = "English",
volume = "6",
pages = "43830--43839",
journal = "IEEE Access",
issn = "2169-3536",
publisher = "Institute of Electrical and Electronics Engineers",

}

TY - JOUR

T1 - Cooperative Correlational and Discriminative Ensemble Classifier Learning for Early Dementia Diagnosis Using Morphological Brain Multiplexes

AU - Raeper, Rory

AU - Lisowska, Anna

AU - Rekik, Islem

PY - 2018/8/28

Y1 - 2018/8/28

N2 - Dementia alters the brain wiring on different levels. However, these changes might be subtle particularly in patients with early mild cognitive impairment (eMCI). Hence, developing accurate diagnostic techniques for eMCI identification is critical for early intervention to prevent the onset of Alzheimer’s Disease (AD). There is a large body of machine-learning based research developed for classifying different brain states (e.g., AD vs MCI) using neuroimaging data. These works can be fundamentally grouped into two categories. The first one uses correlational methods, such as canonical correlation analysis (CCA) and its variants, with the aim to identify most correlated features for diagnosis. The second one includes discriminative methods, such as feature selection methods and linear discriminative analysis (LDA) and its variants to identify brain features that discriminate between two brain states. However, existing methods examine these correlational and discriminative brain data independently, which overlooks the complementary information provided by both techniques, which could prove to be useful in data classification tasks. On the other hand, how early dementia affects cortical brain connections in morphology remains largely unexplored. To address these limitations, we propose a cooperative correlational and discriminative ensemble learning framework for eMCI diagnosis that leverages a brain network representation from multiple morphological networks, each derived from the cortical surface. Specifically, we devise ‘the shallow convolutional brain multiplex’ (SCBM), which encodes both region-to-region and network-to-network relationships. Then, we represent each individual brain using a set of SCBMs, which are used to train an ensemble of CCA-SVM and LDA-based classifiers, cooperating to output the label for a new testing subject. Overall, our framework outperformed several state-of-the-art methods including independent correlational and discriminative methods.

AB - Dementia alters the brain wiring on different levels. However, these changes might be subtle particularly in patients with early mild cognitive impairment (eMCI). Hence, developing accurate diagnostic techniques for eMCI identification is critical for early intervention to prevent the onset of Alzheimer’s Disease (AD). There is a large body of machine-learning based research developed for classifying different brain states (e.g., AD vs MCI) using neuroimaging data. These works can be fundamentally grouped into two categories. The first one uses correlational methods, such as canonical correlation analysis (CCA) and its variants, with the aim to identify most correlated features for diagnosis. The second one includes discriminative methods, such as feature selection methods and linear discriminative analysis (LDA) and its variants to identify brain features that discriminate between two brain states. However, existing methods examine these correlational and discriminative brain data independently, which overlooks the complementary information provided by both techniques, which could prove to be useful in data classification tasks. On the other hand, how early dementia affects cortical brain connections in morphology remains largely unexplored. To address these limitations, we propose a cooperative correlational and discriminative ensemble learning framework for eMCI diagnosis that leverages a brain network representation from multiple morphological networks, each derived from the cortical surface. Specifically, we devise ‘the shallow convolutional brain multiplex’ (SCBM), which encodes both region-to-region and network-to-network relationships. Then, we represent each individual brain using a set of SCBMs, which are used to train an ensemble of CCA-SVM and LDA-based classifiers, cooperating to output the label for a new testing subject. Overall, our framework outperformed several state-of-the-art methods including independent correlational and discriminative methods.

KW - Morphological Brian Network

KW - Multi-View Brain Data

KW - Canonical Correlation Analysis

KW - Discriminative Methds

KW - Linear Discriminant Analysis

KW - Ensemble Classifier

KW - Brain Multiplex

U2 - 10.1109/ACCESS.2018.2863657

DO - 10.1109/ACCESS.2018.2863657

M3 - Article

VL - 6

SP - 43830

EP - 43839

JO - IEEE Access

T2 - IEEE Access

JF - IEEE Access

SN - 2169-3536

ER -