Ensemble sparse classification of Alzheimer's disease

Manhua Liu, Daoqiang Zhang, Dinggang Shen

Research output: Contribution to journalArticle

174 Citations (Scopus)

Abstract

The high-dimensional pattern classification methods, e.g., support vector machines (SVM), have been widely investigated for analysis of structural and functional brain images (such as magnetic resonance imaging (MRI)) to assist the diagnosis of Alzheimer's disease (AD) including its prodromal stage, i.e., mild cognitive impairment (MCI). Most existing classification methods extract features from neuroimaging data and then construct a single classifier to perform classification. However, due to noise and small sample size of neuroimaging data, it is challenging to train only a global classifier that can be robust enough to achieve good classification performance. In this paper, instead of building a single global classifier, we propose a local patch-based subspace ensemble method which builds multiple individual classifiers based on different subsets of local patches and then combines them for more accurate and robust classification. Specifically, to capture the local spatial consistency, each brain image is partitioned into a number of local patches and a subset of patches is randomly selected from the patch pool to build a weak classifier. Here, the sparse representation-based classifier (SRC) method, which has shown to be effective for classification of image data (e.g., face), is used to construct each weak classifier. Then, multiple weak classifiers are combined to make the final decision. We evaluate our method on 652 subjects (including 198 AD patients, 225 MCI and 229 normal controls) from Alzheimer's Disease Neuroimaging Initiative (ADNI) database using MR images. The experimental results show that our method achieves an accuracy of 90.8% and an area under the ROC curve (AUC) of 94.86% for AD classification and an accuracy of 87.85% and an AUC of 92.90% for MCI classification, respectively, demonstrating a very promising performance of our method compared with the state-of-the-art methods for AD/MCI classification using MR images.

Original languageEnglish
Pages (from-to)1106-1116
Number of pages11
JournalNeuroImage
Volume60
Issue number2
DOIs
Publication statusPublished - 2012 Apr 2
Externally publishedYes

Fingerprint

Alzheimer Disease
Neuroimaging
ROC Curve
Area Under Curve
Prodromal Symptoms
Brain
Sample Size
Noise
Magnetic Resonance Imaging
Databases
Cognitive Dysfunction

Keywords

  • AD diagnosis
  • Local patch
  • Random subspace ensemble
  • Sparse representation-based classifier (SRC)

ASJC Scopus subject areas

  • Cognitive Neuroscience
  • Neurology

Cite this

Ensemble sparse classification of Alzheimer's disease. / Liu, Manhua; Zhang, Daoqiang; Shen, Dinggang.

In: NeuroImage, Vol. 60, No. 2, 02.04.2012, p. 1106-1116.

Research output: Contribution to journalArticle

Liu, Manhua ; Zhang, Daoqiang ; Shen, Dinggang. / Ensemble sparse classification of Alzheimer's disease. In: NeuroImage. 2012 ; Vol. 60, No. 2. pp. 1106-1116.
@article{b3b5326e9555477ea38ab25bb85ad77c,
title = "Ensemble sparse classification of Alzheimer's disease",
abstract = "The high-dimensional pattern classification methods, e.g., support vector machines (SVM), have been widely investigated for analysis of structural and functional brain images (such as magnetic resonance imaging (MRI)) to assist the diagnosis of Alzheimer's disease (AD) including its prodromal stage, i.e., mild cognitive impairment (MCI). Most existing classification methods extract features from neuroimaging data and then construct a single classifier to perform classification. However, due to noise and small sample size of neuroimaging data, it is challenging to train only a global classifier that can be robust enough to achieve good classification performance. In this paper, instead of building a single global classifier, we propose a local patch-based subspace ensemble method which builds multiple individual classifiers based on different subsets of local patches and then combines them for more accurate and robust classification. Specifically, to capture the local spatial consistency, each brain image is partitioned into a number of local patches and a subset of patches is randomly selected from the patch pool to build a weak classifier. Here, the sparse representation-based classifier (SRC) method, which has shown to be effective for classification of image data (e.g., face), is used to construct each weak classifier. Then, multiple weak classifiers are combined to make the final decision. We evaluate our method on 652 subjects (including 198 AD patients, 225 MCI and 229 normal controls) from Alzheimer's Disease Neuroimaging Initiative (ADNI) database using MR images. The experimental results show that our method achieves an accuracy of 90.8{\%} and an area under the ROC curve (AUC) of 94.86{\%} for AD classification and an accuracy of 87.85{\%} and an AUC of 92.90{\%} for MCI classification, respectively, demonstrating a very promising performance of our method compared with the state-of-the-art methods for AD/MCI classification using MR images.",
keywords = "AD diagnosis, Local patch, Random subspace ensemble, Sparse representation-based classifier (SRC)",
author = "Manhua Liu and Daoqiang Zhang and Dinggang Shen",
year = "2012",
month = "4",
day = "2",
doi = "10.1016/j.neuroimage.2012.01.055",
language = "English",
volume = "60",
pages = "1106--1116",
journal = "NeuroImage",
issn = "1053-8119",
publisher = "Academic Press Inc.",
number = "2",

}

TY - JOUR

T1 - Ensemble sparse classification of Alzheimer's disease

AU - Liu, Manhua

AU - Zhang, Daoqiang

AU - Shen, Dinggang

PY - 2012/4/2

Y1 - 2012/4/2

N2 - The high-dimensional pattern classification methods, e.g., support vector machines (SVM), have been widely investigated for analysis of structural and functional brain images (such as magnetic resonance imaging (MRI)) to assist the diagnosis of Alzheimer's disease (AD) including its prodromal stage, i.e., mild cognitive impairment (MCI). Most existing classification methods extract features from neuroimaging data and then construct a single classifier to perform classification. However, due to noise and small sample size of neuroimaging data, it is challenging to train only a global classifier that can be robust enough to achieve good classification performance. In this paper, instead of building a single global classifier, we propose a local patch-based subspace ensemble method which builds multiple individual classifiers based on different subsets of local patches and then combines them for more accurate and robust classification. Specifically, to capture the local spatial consistency, each brain image is partitioned into a number of local patches and a subset of patches is randomly selected from the patch pool to build a weak classifier. Here, the sparse representation-based classifier (SRC) method, which has shown to be effective for classification of image data (e.g., face), is used to construct each weak classifier. Then, multiple weak classifiers are combined to make the final decision. We evaluate our method on 652 subjects (including 198 AD patients, 225 MCI and 229 normal controls) from Alzheimer's Disease Neuroimaging Initiative (ADNI) database using MR images. The experimental results show that our method achieves an accuracy of 90.8% and an area under the ROC curve (AUC) of 94.86% for AD classification and an accuracy of 87.85% and an AUC of 92.90% for MCI classification, respectively, demonstrating a very promising performance of our method compared with the state-of-the-art methods for AD/MCI classification using MR images.

AB - The high-dimensional pattern classification methods, e.g., support vector machines (SVM), have been widely investigated for analysis of structural and functional brain images (such as magnetic resonance imaging (MRI)) to assist the diagnosis of Alzheimer's disease (AD) including its prodromal stage, i.e., mild cognitive impairment (MCI). Most existing classification methods extract features from neuroimaging data and then construct a single classifier to perform classification. However, due to noise and small sample size of neuroimaging data, it is challenging to train only a global classifier that can be robust enough to achieve good classification performance. In this paper, instead of building a single global classifier, we propose a local patch-based subspace ensemble method which builds multiple individual classifiers based on different subsets of local patches and then combines them for more accurate and robust classification. Specifically, to capture the local spatial consistency, each brain image is partitioned into a number of local patches and a subset of patches is randomly selected from the patch pool to build a weak classifier. Here, the sparse representation-based classifier (SRC) method, which has shown to be effective for classification of image data (e.g., face), is used to construct each weak classifier. Then, multiple weak classifiers are combined to make the final decision. We evaluate our method on 652 subjects (including 198 AD patients, 225 MCI and 229 normal controls) from Alzheimer's Disease Neuroimaging Initiative (ADNI) database using MR images. The experimental results show that our method achieves an accuracy of 90.8% and an area under the ROC curve (AUC) of 94.86% for AD classification and an accuracy of 87.85% and an AUC of 92.90% for MCI classification, respectively, demonstrating a very promising performance of our method compared with the state-of-the-art methods for AD/MCI classification using MR images.

KW - AD diagnosis

KW - Local patch

KW - Random subspace ensemble

KW - Sparse representation-based classifier (SRC)

UR - http://www.scopus.com/inward/record.url?scp=84862778147&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84862778147&partnerID=8YFLogxK

U2 - 10.1016/j.neuroimage.2012.01.055

DO - 10.1016/j.neuroimage.2012.01.055

M3 - Article

C2 - 22270352

AN - SCOPUS:84862778147

VL - 60

SP - 1106

EP - 1116

JO - NeuroImage

JF - NeuroImage

SN - 1053-8119

IS - 2

ER -