Large margin local estimate with applications to medical image classification

Yang Song, Weidong Cai, Heng Huang, Yun Zhou, David Dagan Feng, Yue Wang, Michael J. Fulham, Mei Chen

Research output: Contribution to journalArticle

Abstract

Medical images usually exhibit large intra-class variation and inter-class ambiguity in the feature space, which could affect classification accuracy. To tackle this issue, we propose a new Large Margin Local Estimate (LMLE) classification model with sub-categorization based sparse representation. We first sub-categorize the reference sets of different classes into multiple clusters, to reduce feature variation within each subcategory compared to the entire reference set. Local estimates are generated for the test image using sparse representation with reference subcategories as the dictionaries. The similarity between the test image and each class is then computed by fusing the distances with the local estimates in a learning-based large margin aggregation construct to alleviate the problem of inter-class ambiguity. The derived similarities are finally used to determine the class label. We demonstrate that our LMLE model is generally applicable to different imaging modalities, and applied it to three tasks: interstitial lung disease (ILD) classification on high-resolution computed tomography (HRCT) images, phenotype binary classification and continuous regression on brain magnetic resonance (MR) imaging. Our experimental results show statistically significant performance improvements over existing popular classifiers.

Original languageEnglish (US)
Article number7014242
Pages (from-to)1362-1377
Number of pages16
JournalIEEE Transactions on Medical Imaging
Volume34
Issue number6
DOIs
StatePublished - Jun 1 2015

Fingerprint

Image classification
Imaging techniques
Pulmonary diseases
Binary images
Interstitial Lung Diseases
Magnetic resonance
Glossaries
Tomography
Labels
Brain
Classifiers
Agglomeration
Magnetic Resonance Imaging
Learning
Phenotype

Keywords

  • Large margin fusion
  • medical image classification
  • sparse representation
  • sub-categorization

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Computer Science Applications
  • Radiological and Ultrasound Technology
  • Software

Cite this

Song, Y., Cai, W., Huang, H., Zhou, Y., Feng, D. D., Wang, Y., ... Chen, M. (2015). Large margin local estimate with applications to medical image classification. IEEE Transactions on Medical Imaging, 34(6), 1362-1377. [7014242]. https://doi.org/10.1109/TMI.2015.2393954

Large margin local estimate with applications to medical image classification. / Song, Yang; Cai, Weidong; Huang, Heng; Zhou, Yun; Feng, David Dagan; Wang, Yue; Fulham, Michael J.; Chen, Mei.

In: IEEE Transactions on Medical Imaging, Vol. 34, No. 6, 7014242, 01.06.2015, p. 1362-1377.

Research output: Contribution to journalArticle

Song, Y, Cai, W, Huang, H, Zhou, Y, Feng, DD, Wang, Y, Fulham, MJ & Chen, M 2015, 'Large margin local estimate with applications to medical image classification', IEEE Transactions on Medical Imaging, vol. 34, no. 6, 7014242, pp. 1362-1377. https://doi.org/10.1109/TMI.2015.2393954
Song, Yang ; Cai, Weidong ; Huang, Heng ; Zhou, Yun ; Feng, David Dagan ; Wang, Yue ; Fulham, Michael J. ; Chen, Mei. / Large margin local estimate with applications to medical image classification. In: IEEE Transactions on Medical Imaging. 2015 ; Vol. 34, No. 6. pp. 1362-1377.
@article{acd603ec5e2442e0b5463efdc8703365,
title = "Large margin local estimate with applications to medical image classification",
abstract = "Medical images usually exhibit large intra-class variation and inter-class ambiguity in the feature space, which could affect classification accuracy. To tackle this issue, we propose a new Large Margin Local Estimate (LMLE) classification model with sub-categorization based sparse representation. We first sub-categorize the reference sets of different classes into multiple clusters, to reduce feature variation within each subcategory compared to the entire reference set. Local estimates are generated for the test image using sparse representation with reference subcategories as the dictionaries. The similarity between the test image and each class is then computed by fusing the distances with the local estimates in a learning-based large margin aggregation construct to alleviate the problem of inter-class ambiguity. The derived similarities are finally used to determine the class label. We demonstrate that our LMLE model is generally applicable to different imaging modalities, and applied it to three tasks: interstitial lung disease (ILD) classification on high-resolution computed tomography (HRCT) images, phenotype binary classification and continuous regression on brain magnetic resonance (MR) imaging. Our experimental results show statistically significant performance improvements over existing popular classifiers.",
keywords = "Large margin fusion, medical image classification, sparse representation, sub-categorization",
author = "Yang Song and Weidong Cai and Heng Huang and Yun Zhou and Feng, {David Dagan} and Yue Wang and Fulham, {Michael J.} and Mei Chen",
year = "2015",
month = "6",
day = "1",
doi = "10.1109/TMI.2015.2393954",
language = "English (US)",
volume = "34",
pages = "1362--1377",
journal = "IEEE Transactions on Medical Imaging",
issn = "0278-0062",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "6",

}

TY - JOUR

T1 - Large margin local estimate with applications to medical image classification

AU - Song, Yang

AU - Cai, Weidong

AU - Huang, Heng

AU - Zhou, Yun

AU - Feng, David Dagan

AU - Wang, Yue

AU - Fulham, Michael J.

AU - Chen, Mei

PY - 2015/6/1

Y1 - 2015/6/1

N2 - Medical images usually exhibit large intra-class variation and inter-class ambiguity in the feature space, which could affect classification accuracy. To tackle this issue, we propose a new Large Margin Local Estimate (LMLE) classification model with sub-categorization based sparse representation. We first sub-categorize the reference sets of different classes into multiple clusters, to reduce feature variation within each subcategory compared to the entire reference set. Local estimates are generated for the test image using sparse representation with reference subcategories as the dictionaries. The similarity between the test image and each class is then computed by fusing the distances with the local estimates in a learning-based large margin aggregation construct to alleviate the problem of inter-class ambiguity. The derived similarities are finally used to determine the class label. We demonstrate that our LMLE model is generally applicable to different imaging modalities, and applied it to three tasks: interstitial lung disease (ILD) classification on high-resolution computed tomography (HRCT) images, phenotype binary classification and continuous regression on brain magnetic resonance (MR) imaging. Our experimental results show statistically significant performance improvements over existing popular classifiers.

AB - Medical images usually exhibit large intra-class variation and inter-class ambiguity in the feature space, which could affect classification accuracy. To tackle this issue, we propose a new Large Margin Local Estimate (LMLE) classification model with sub-categorization based sparse representation. We first sub-categorize the reference sets of different classes into multiple clusters, to reduce feature variation within each subcategory compared to the entire reference set. Local estimates are generated for the test image using sparse representation with reference subcategories as the dictionaries. The similarity between the test image and each class is then computed by fusing the distances with the local estimates in a learning-based large margin aggregation construct to alleviate the problem of inter-class ambiguity. The derived similarities are finally used to determine the class label. We demonstrate that our LMLE model is generally applicable to different imaging modalities, and applied it to three tasks: interstitial lung disease (ILD) classification on high-resolution computed tomography (HRCT) images, phenotype binary classification and continuous regression on brain magnetic resonance (MR) imaging. Our experimental results show statistically significant performance improvements over existing popular classifiers.

KW - Large margin fusion

KW - medical image classification

KW - sparse representation

KW - sub-categorization

UR - http://www.scopus.com/inward/record.url?scp=84930947629&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84930947629&partnerID=8YFLogxK

U2 - 10.1109/TMI.2015.2393954

DO - 10.1109/TMI.2015.2393954

M3 - Article

VL - 34

SP - 1362

EP - 1377

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

SN - 0278-0062

IS - 6

M1 - 7014242

ER -