Generative-discriminative basis learning for medical imaging

Nematollah K. Batmanghelich, Ben Taskar, Christos Davatzikos

Research output: Contribution to journalArticle

Abstract

This paper presents a novel dimensionality reduction method for classification in medical imaging. The goal is to transform very high-dimensional input (typically, millions of voxels) to a low-dimensional representation (small number of constructed features) that preserves discriminative signal and is clinically interpretable. We formulate the task as a constrained optimization problem that combines generative and discriminative objectives and show how to extend it to the semi-supervised learning (SSL) setting. We propose a novel large-scale algorithm to solve the resulting optimization problem. In the fully supervised case, we demonstrate accuracy rates that are better than or comparable to state-of-the-art algorithms on several datasets while producing a representation of the group difference that is consistent with prior clinical reports. Effectiveness of the proposed algorithm for SSL is evaluated with both benchmark and medical imaging datasets. In the benchmark datasets, the results are better than or comparable to the state-of-the-art methods for SSL. For evaluation of the SSL setting in medical datasets, we use images of subjects with mild cognitive impairment (MCI), which is believed to be a precursor to Alzheimer's disease (AD), as unlabeled data. AD subjects and normal control (NC) subjects are used as labeled data, and we try to predict conversion from MCI to AD on follow-up. The semi-supervised extension of this method not only improves the generalization accuracy for the labeled data (AD/NC) slightly but is also able to predict subjects which are likely to converge to AD.

Original languageEnglish (US)
Article number5961630
Pages (from-to)51-69
Number of pages19
JournalIEEE Transactions on Medical Imaging
Volume31
Issue number1
DOIs
StatePublished - Jan 2012
Externally publishedYes

Fingerprint

Medical imaging
Diagnostic Imaging
Supervised learning
Alzheimer Disease
Learning
Benchmarking
Constrained optimization
Datasets
Supervised Machine Learning
Cognitive Dysfunction

Keywords

  • Basis learning
  • classification
  • feature construction
  • generative-discriminative learning
  • machine learning
  • matrix factorization
  • morphological pattern analysis
  • optimization
  • semi-supervised learning
  • sparsity

ASJC Scopus subject areas

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

Cite this

Generative-discriminative basis learning for medical imaging. / Batmanghelich, Nematollah K.; Taskar, Ben; Davatzikos, Christos.

In: IEEE Transactions on Medical Imaging, Vol. 31, No. 1, 5961630, 01.2012, p. 51-69.

Research output: Contribution to journalArticle

Batmanghelich, NK, Taskar, B & Davatzikos, C 2012, 'Generative-discriminative basis learning for medical imaging', IEEE Transactions on Medical Imaging, vol. 31, no. 1, 5961630, pp. 51-69. https://doi.org/10.1109/TMI.2011.2162961
Batmanghelich, Nematollah K. ; Taskar, Ben ; Davatzikos, Christos. / Generative-discriminative basis learning for medical imaging. In: IEEE Transactions on Medical Imaging. 2012 ; Vol. 31, No. 1. pp. 51-69.
@article{3c44cf07bc774ceba9c33bbc6cba93dd,
title = "Generative-discriminative basis learning for medical imaging",
abstract = "This paper presents a novel dimensionality reduction method for classification in medical imaging. The goal is to transform very high-dimensional input (typically, millions of voxels) to a low-dimensional representation (small number of constructed features) that preserves discriminative signal and is clinically interpretable. We formulate the task as a constrained optimization problem that combines generative and discriminative objectives and show how to extend it to the semi-supervised learning (SSL) setting. We propose a novel large-scale algorithm to solve the resulting optimization problem. In the fully supervised case, we demonstrate accuracy rates that are better than or comparable to state-of-the-art algorithms on several datasets while producing a representation of the group difference that is consistent with prior clinical reports. Effectiveness of the proposed algorithm for SSL is evaluated with both benchmark and medical imaging datasets. In the benchmark datasets, the results are better than or comparable to the state-of-the-art methods for SSL. For evaluation of the SSL setting in medical datasets, we use images of subjects with mild cognitive impairment (MCI), which is believed to be a precursor to Alzheimer's disease (AD), as unlabeled data. AD subjects and normal control (NC) subjects are used as labeled data, and we try to predict conversion from MCI to AD on follow-up. The semi-supervised extension of this method not only improves the generalization accuracy for the labeled data (AD/NC) slightly but is also able to predict subjects which are likely to converge to AD.",
keywords = "Basis learning, classification, feature construction, generative-discriminative learning, machine learning, matrix factorization, morphological pattern analysis, optimization, semi-supervised learning, sparsity",
author = "Batmanghelich, {Nematollah K.} and Ben Taskar and Christos Davatzikos",
year = "2012",
month = "1",
doi = "10.1109/TMI.2011.2162961",
language = "English (US)",
volume = "31",
pages = "51--69",
journal = "IEEE Transactions on Medical Imaging",
issn = "0278-0062",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "1",

}

TY - JOUR

T1 - Generative-discriminative basis learning for medical imaging

AU - Batmanghelich, Nematollah K.

AU - Taskar, Ben

AU - Davatzikos, Christos

PY - 2012/1

Y1 - 2012/1

N2 - This paper presents a novel dimensionality reduction method for classification in medical imaging. The goal is to transform very high-dimensional input (typically, millions of voxels) to a low-dimensional representation (small number of constructed features) that preserves discriminative signal and is clinically interpretable. We formulate the task as a constrained optimization problem that combines generative and discriminative objectives and show how to extend it to the semi-supervised learning (SSL) setting. We propose a novel large-scale algorithm to solve the resulting optimization problem. In the fully supervised case, we demonstrate accuracy rates that are better than or comparable to state-of-the-art algorithms on several datasets while producing a representation of the group difference that is consistent with prior clinical reports. Effectiveness of the proposed algorithm for SSL is evaluated with both benchmark and medical imaging datasets. In the benchmark datasets, the results are better than or comparable to the state-of-the-art methods for SSL. For evaluation of the SSL setting in medical datasets, we use images of subjects with mild cognitive impairment (MCI), which is believed to be a precursor to Alzheimer's disease (AD), as unlabeled data. AD subjects and normal control (NC) subjects are used as labeled data, and we try to predict conversion from MCI to AD on follow-up. The semi-supervised extension of this method not only improves the generalization accuracy for the labeled data (AD/NC) slightly but is also able to predict subjects which are likely to converge to AD.

AB - This paper presents a novel dimensionality reduction method for classification in medical imaging. The goal is to transform very high-dimensional input (typically, millions of voxels) to a low-dimensional representation (small number of constructed features) that preserves discriminative signal and is clinically interpretable. We formulate the task as a constrained optimization problem that combines generative and discriminative objectives and show how to extend it to the semi-supervised learning (SSL) setting. We propose a novel large-scale algorithm to solve the resulting optimization problem. In the fully supervised case, we demonstrate accuracy rates that are better than or comparable to state-of-the-art algorithms on several datasets while producing a representation of the group difference that is consistent with prior clinical reports. Effectiveness of the proposed algorithm for SSL is evaluated with both benchmark and medical imaging datasets. In the benchmark datasets, the results are better than or comparable to the state-of-the-art methods for SSL. For evaluation of the SSL setting in medical datasets, we use images of subjects with mild cognitive impairment (MCI), which is believed to be a precursor to Alzheimer's disease (AD), as unlabeled data. AD subjects and normal control (NC) subjects are used as labeled data, and we try to predict conversion from MCI to AD on follow-up. The semi-supervised extension of this method not only improves the generalization accuracy for the labeled data (AD/NC) slightly but is also able to predict subjects which are likely to converge to AD.

KW - Basis learning

KW - classification

KW - feature construction

KW - generative-discriminative learning

KW - machine learning

KW - matrix factorization

KW - morphological pattern analysis

KW - optimization

KW - semi-supervised learning

KW - sparsity

UR - http://www.scopus.com/inward/record.url?scp=84855382390&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84855382390&partnerID=8YFLogxK

U2 - 10.1109/TMI.2011.2162961

DO - 10.1109/TMI.2011.2162961

M3 - Article

C2 - 21791408

AN - SCOPUS:84855382390

VL - 31

SP - 51

EP - 69

JO - IEEE Transactions on Medical Imaging

JF - IEEE Transactions on Medical Imaging

SN - 0278-0062

IS - 1

M1 - 5961630

ER -