Integral Transform Technique for Monitoring Electrocardiograms.

Abstract

The paper applies time domain and filtered integral transform domain pattern recognition techniques to the problem of locating and discriminating between QRS-complexes and premature ventricular contractions in electrocardiograms. The data set consists of sixteen 100-second single lead electrocardiogram records. Sufficient electrocardiogram and pattern recognition theory is included to explain the technique and the results. Three different integral transforms are evaluated: the discrete Fourier, the Walsh, and an accidentally discovered transform similar to the Walsh. Ninety-seven percent of the QRS-complexes and premature ventricular contractions (PVC's) in the records evaluated were located correctly. The three transform techniques classified between 90.8% and 92.5% of the QRS-complexes correctly and 85.6% and 93.2% of the PVC's correctly. (Author)

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1973
Accession Number
AD0769297

Entities

People

  • James H. Reid

Organizations

  • Air Force Institute of Technology

Tags

DTIC Thesaurus Topics

  • Data Sets
  • Diagnostic Techniques (Medicine)
  • Electrocardiography
  • Integral Transforms
  • Integrals
  • Mathematics
  • Monitoring
  • Pattern Recognition
  • Recognition
  • Time Domain

Readers

  • Approximation Theory.
  • Cardiovascular Physiology
  • Regression Analysis.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks