Medical & Biological Engineering & Computing

, Volume 51, Issue 5, pp 513–523

Atherosclerotic plaque tissue characterization in 2D ultrasound longitudinal carotid scans for automated classification: a paradigm for stroke risk assessment

Authors

  • U. Rajendra Acharya
    • Department of Electronics and Computer EngineeringNgee Ann Polytechnic
    • Department of Biomedical Engineering, Faculty of EngineeringUniversity of Malaya
  • Muthu Rama Krishnan Mookiah
    • Department of Electronics and Computer EngineeringNgee Ann Polytechnic
    • Global Biomedical Technologies Inc.
  • David Afonso
    • Department of Electrical and Computer EngineeringInstituto Superior Tecnico
  • Joao Sanches
    • Department of Electrical and Computer EngineeringInstituto Superior Tecnico
  • Shoaib Shafique
    • CorVasc MDs
  • Andrew Nicolaides
    • Vascular Screening and Diagnostic Centre, London, Imperial college
    • Department of Biological SciencesUniversity of Cyprus
  • L. M. Pedro
    • Cardiovascular Institute and the Lisbon University Medical School, Hospital de SantaMaria
  • J. Fernandes e Fernandes
    • Cardiovascular Institute and the Lisbon University Medical School, Hospital de SantaMaria
  • Jasjit S. Suri
    • Global Biomedical Technologies Inc.
    • Department of Biomedical EngineeringIdaho State University
    • Diagnostic and Monitoring DivisionAtheroPoint(TM) LLC Roseville
Original Article

DOI: 10.1007/s11517-012-1019-0

Cite this article as:
Acharya, U.R., Mookiah, M.R.K., Vinitha Sree, S. et al. Med Biol Eng Comput (2013) 51: 513. doi:10.1007/s11517-012-1019-0

Abstract

In the case of carotid atherosclerosis, to avoid unnecessary surgeries in asymptomatic patients, it is necessary to develop a technique to effectively differentiate symptomatic and asymptomatic plaques. In this paper, we have presented a data mining framework that characterizes the textural differences in these two classes using several grayscale features based on a novel combination of trace transform and fuzzy texture. The features extracted from the delineated plaque regions in B-mode ultrasound images were used to train several classifiers in order to prepare them for classification of new test plaques. Our CAD system was evaluated using two different databases consisting of 146 (44 symptomatic to 102 asymptomatic) and 346 (196 symptomatic and 150 asymptomatic) images. Both these databases differ in the way the ground truth was determined. We obtained classification accuracies of 93.1 and 85.3 %, respectively. The techniques are low cost, easily implementable, objective, and non-invasive. For more objective analysis, we have also developed novel integrated indices using a combination of significant features.

Keywords

AtherosclerosisCarotid ultrasoundClassificationFuzzy textureIndexPlaque characterizationTrace transform

Copyright information

© International Federation for Medical and Biological Engineering 2013