Title:
Novel method of digitization of electrocardiogram signals
Novel method of digitization of electrocardiogram signals
Author(s)
Ganesh, Shambavi
Advisor(s)
Bhatti, Pamela T.
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
The objective of the proposed research is to implement a novel digitization tool which extracts electrocardiogram (ECG) signals, as well as patient demographic information from paper electrocardiography records. A MATLAB (MathWorks: Natick, MA) based graphical user interface (GUI) application is proposed wherein the electrocardiogram (ECG) signals are digitized based on an algorithm built upon previous work by Ravichandran et al. [7]. The existing algorithm carried out Optical Character Recognition (OCR) for the extraction of the demographic information from the record. In addition to this existing function, the proposed research aims to implement OCR on the ECG character lead names in order to further enhance the accuracy of the digitized ECG signals. The proposed algorithm is subjected to a reader study conducted at Emory University. In order to measure the correlation between the digitized ECG signals and the paper ECG records, the reader study aimed to measure the following parameters: QT (the time measured from the QRS complex to the end of T-wave), QRS (the duration measured along the length of the QRS complex), PR (the time measured from the beginning of the P-wave to that of the QRS complex), QTc (the increase in duration of action potential of signal) and RR (the time measured between 2 consecutive R-R peaks). From the intra-observer and inter-observer correlation values between the digitized and paper ECG records ranged between 0.2557 to 0.9371. Although the RR intervals is found to be statistically significant, the QT and QTc intervals are not. The kappa statistic ranged between -0.4886 to 0.8742, for the intra-observer measurements. For the inter-observer measurements, the kappa statistic ranged between 0.1722 and 0.8216.
Sponsor
Date Issued
2020-04-28
Extent
Resource Type
Text
Resource Subtype
Thesis