Bayesian classification of myocardial excitation abnormality using magnetocardiogram maps for mass screening

Yumie Ono, A. Ishiyama, N. Kasai, S. Yamada, K. On, S. Watanabe, I. Yamaguchi, T. Miyashita, K. Tsukada

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

We propose a novel classification method based on the Bayes rule to utilize the magnetocardiogram (MCG) in noninvasive mass screening. The cardiac excitation is directly tracked by maps of the MCG field generated by myocardial excitation current through the excited wave front. To adopt the characteristics of the excited wave fronts as a parameter for the Bayes theorem, we developed a parameterization procedure that consists of a two-dimensional wavelet approximation and a cluster analysis of magnetic field maps. With the parameter determined by this procedure, the probability of a subject to belong to a disease group or to the normal group is estimated by the Bayes theorem. The subject is classified into the group of the highest probability. We applied the proposed method to ST-T period of MCG data of 6 old myocardial infarction (OMI) patients and 15 normal controls. The method showed sensitivity of 83%; specificity, 100%; positive predictive value, 100%; and negative predictive value, 94% in the classification of OMI patients and normal controls. The processing time is less than 5 seconds per one subject. It suggests a possible application of the proposed method in mass screening of abnormal MCG patterns.

Original languageEnglish
JournalNeurology and Clinical Neurophysiology
Volume2004
Publication statusPublished - 2004 Jan 1

Keywords

  • Bayes criterion
  • Excited wave front
  • Gaussian mixture modeling
  • Magnetocardiogram
  • Myocardial infarction
  • Screening
  • Wavelet approximation

ASJC Scopus subject areas

  • Neuroscience(all)

Fingerprint Dive into the research topics of 'Bayesian classification of myocardial excitation abnormality using magnetocardiogram maps for mass screening'. Together they form a unique fingerprint.

Cite this