The following information was submitted:
Transactions: INTERNATIONAL JOURNAL of CIRCUITS, SYSTEMS and SIGNAL PROCESSING
Transactions ID Number: 20-456
Full Name: Gavin Shorten
Position: Student
Age: ON
Sex: Male
Address: The Printing House, Trinity College Dublin, Dublin 2.
Country: IRELAND
Tel:
Tel prefix:
Fax:
E-mail address: shorteg@tcd.ie
Other E-mails:
Title of the Paper: The Application of Dynamic Time Warping to Measure the Accuracy of ECG Compression
Authors as they appear in the Paper: Gavin P. Shorten and Martin J. Burke
Email addresses of all the authors: shorteg@tcd.ie,martin.burke@tcd.ie
Number of paper pages: 9
Abstract: Many different methods of ECG compression have been suggested over the last number of decades. They are typically classed into to three distinct groups â€" Direct Data, Parameter Extraction and Transform methods. The metric most frequently used to differentiate between the accuracy of the different types of compression is a percentage root-mean-square difference (PRD) calculation versus compression ratio, despite the accuracy of such a method having been acknowledged as greatly limited. In this article PRD calculation and an improved partial PRD difference method are investigated and their significant shortcomings highlighted. Dynamic time warping is presented as a method of quantifying the approximation error which may be present but goes undetected by the percentage PRD calculation due to approximation of the ECG. Dynamic time warping provides a significantly more accurate metric for comparing compression algorithms and their respective accuracies. It allows for !
detailed comparison of differing approximation methods and variations of the same approximation method, not possible using the RMS difference versus compression ratio. Its usefulness is fully investigated by comparing several direct data compression algorithms, including a novel threshold variation of the scan along polygonal approximation technique. Results provided demonstrate possible misdiagnosis of cardiac conditions resulting from the limitation of the RMS difference versus compression ratio metric and the benefits of the application of dynamic time warping in examining the accuracy of different compression techniques.
Keywords: ECG Compression, Dynamic Time Warping, Approximation Error, Scan Along Polygonal Approximation, Error Threshold, Pattern Recognition, Signal Processing.
EXTENSION of the file: .pdf
Special (Invited) Session: A Novel Approach in Testing the Accuracy of ECG Compression using Partial Percentage RMS Difference and Dynamic Time Warping
Organizer of the Session: ISPRA 2011 Cambridge Article number 650-174.
How Did you learn about congress:
IP ADDRESS: 86.42.197.253
