A Low Complexity Architecture for Online On-chip Detection and Identification of f-QRS Feature for Remote Personalized Health Care Applications

Abstract

This paper introduces a novel low complexity highly accurate on-chip architecture for the detection of fragmented QRS (f-QRS) feature including notches and local extrema in the QRS complexes and subsequently identifies its various morphologies (Notched S, rsR', RsR' without elevation etc.) under the real-time environment targeting remote personalized health care. The proposed architecture uses the outcome of recently proposed Hybrid feature extraction algorithm (HFEA) [1] Level 3 detailed coefficients and detects and identifies the fragmentation feature from the QRS complex based on the criteria of the positions, and the magnitudes of the extrema (maxima and minima) and notches from the wavelet coefficients with no extra cost in terms of arithmetic complexity. To verify the proposed architecture 100 patients were randomly selected from the MIT-BIH PhysioNet PTB database and their ECG was examined by two experienced cardiologists individually and the results were compared with those obtained from the architecture output wherein we have achieved 95 % diagnostic matching.