Multi-lead Data Extraction Method of ECG Waveform Based on Mobile Device Application

Abstract

Based on its portability and immediacy, mobile device have evolved and led to many innovations in medical domains. For electrocardiograms, the quickest and most intuitive way to store them is to take digital pictures. However, there are distortion and noise interference in real scene photographs inevitably. In this paper, a method is proposed to transform the preliminary digital image achieved by camera into the binarized multi-lead ECG curve individually. The vertex searching algorithm is designed to determine the location of the target paper against an intricate background. The vertex coordinates automatically detected are used in reverse perspective transformation to correct geometric distortion. A robust improved OTSU algorithm is developed to separate foreground curve from background noise and grid. In order to divide the lead curve region, the template matching technology is introduced to the recognition and the region segmentation of each lead information. Finally, a complement operation on the broken curve is performed and the pure and integrated ECG curve which suffices 1-D digitalization is obtained. The entire algorithm can standardize ECG from different sources and solve the problem of region sharing caused by data encryption.