Publication detail

Clinical Accuracy QRS Detector with Automatic Parameter Adjustment in an Autonomous, Real-Time Physiologic Monitor

PINTO, S. FELTON, C. SMITAL, L. GILBERT, B. HOLMES, D. HAIDER, C.

Original Title

Clinical Accuracy QRS Detector with Automatic Parameter Adjustment in an Autonomous, Real-Time Physiologic Monitor

English Title

Clinical Accuracy QRS Detector with Automatic Parameter Adjustment in an Autonomous, Real-Time Physiologic Monitor

Type

conference paper

Language

en

Original Abstract

This paper presents a computationally and temporal data-compact QRS complex detection algorithm useful in embedded real-time electrocardiogram (ECG) waveform analysis. The aim of the compact algorithms is to provide high sensitivity and specificity, i.e. diagnostically useful QRS waveform detection, in a continuous ambulatory monitor setting. The proposed detector uses a multi-level approach: QRS highlighting by means of a Truncated Discrete Time Stockwell Transform (TDTST), peak discrimination, and a trained Neural Network to reduce the number of false positive QRS detections. An optimization method is presented that automatically adjust the detector’s parameters to minimize the computational cost. Results demonstrate that the compact TDTST algorithm exhibits high QRS detection accuracy, an error rate of 0.31%, and remains applicable to real-time embedded physiologic ambulatory monitors.

English abstract

This paper presents a computationally and temporal data-compact QRS complex detection algorithm useful in embedded real-time electrocardiogram (ECG) waveform analysis. The aim of the compact algorithms is to provide high sensitivity and specificity, i.e. diagnostically useful QRS waveform detection, in a continuous ambulatory monitor setting. The proposed detector uses a multi-level approach: QRS highlighting by means of a Truncated Discrete Time Stockwell Transform (TDTST), peak discrimination, and a trained Neural Network to reduce the number of false positive QRS detections. An optimization method is presented that automatically adjust the detector’s parameters to minimize the computational cost. Results demonstrate that the compact TDTST algorithm exhibits high QRS detection accuracy, an error rate of 0.31%, and remains applicable to real-time embedded physiologic ambulatory monitors.

Keywords

Compact algorithm, Stockwell transform, Realtime, Physiologic monitoring, QRS detection, Embedded

Released

14.11.2017

Publisher

IEEE

Location

Montreal, QC, Kanada

ISBN

978-1-5090-5990-4

Book

2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP)

Pages from

1005

Pages to

1009

Pages count

5

URL

BibTex


@inproceedings{BUT143750,
  author="Samuel Cerqueira {Pinto} and Christopher L. {Felton} and Lukáš {Smital} and Barry {Gilbert} and David {Holmes} and Clifton {Haider}",
  title="Clinical Accuracy QRS Detector with Automatic Parameter Adjustment in an Autonomous, Real-Time Physiologic Monitor",
  annote="This paper presents a computationally and temporal data-compact QRS complex detection algorithm useful in embedded real-time electrocardiogram (ECG) waveform analysis. The aim of the compact algorithms is to provide high sensitivity and specificity, i.e. diagnostically useful QRS waveform detection, in a continuous ambulatory monitor setting. The proposed detector uses a multi-level approach: QRS highlighting by means of a Truncated Discrete Time Stockwell Transform (TDTST), peak discrimination, and a trained Neural Network to reduce the number of false positive QRS detections. An optimization method is presented that automatically adjust the detector’s parameters to minimize the computational cost. Results demonstrate that the compact TDTST algorithm exhibits high QRS detection accuracy, an error rate of 0.31%, and remains applicable to real-time embedded physiologic ambulatory monitors.",
  address="IEEE",
  booktitle="2017 IEEE Global Conference on Signal and Information Processing (GlobalSIP)",
  chapter="143750",
  doi="10.1109/GlobalSIP.2017.8309112",
  howpublished="online",
  institution="IEEE",
  year="2017",
  month="november",
  pages="1005--1009",
  publisher="IEEE",
  type="conference paper"
}