Session S63.1

Assessing QT Interval Hysteresis from Standard 12-Lead ECG Measurements

V Varadarajan *, VN Polotski, CP Danford, JM Starobin

Mediwave Star Technology Inc.
Greensboro, NC, USA

The amount of QT-RR interval hysteresis accumulated during an exercise stress test between load and recovery phases is associated with the degree of myocardial ischemia. Therefore the evaluation of hysteresis from standard 12-lead ECG (12-SL) is an important practical modality for the assessment of severity of coronary artery disease. Commercial QT and RR waveform analyzers are regularly associated with interval detection artifacts and inconsistencies in one or more processed leads. These artifacts render ambiguity in estimates of hysteresis. We present a signal processing based technique which quantifies the level of artifacts in different leads of the 12-SL thereby eliminating the need to analyze faulty data from these leads.
In order to calculate the magnitude of QT-RR hysteresis we model heart rate dependent QT interval variations (QT interval dynamics) as the output of an unknown finite linear filter w with M parameters. This filter is designed to estimate the components of QT intervals that are correlated to the recent history of RR intervals. The filter length M determines the number of preceding RR intervals used in each estimate. The optimization of the filter parameters is based on minimizing error between the estimates and recordings using a mean square error (MSE) algorithm. Using an empirical MSE based threshold we identify a subset of leads with consistent QT interval measurements and few artifacts. A median estimate for hysteresis is computed from this subset.
QT and RR interval data from 20 patients in an ongoing pilot study were measured using the GE CASE waveform analyzer. For each patient QT and RR intervals were recorded in all 12 leads. Implementation of the MSE algorithm allowed us to identify on average 3-5 leads with a reduced fraction of artifacts and reliable measures of hysteresis. In addition to processing of QT-RR hysteresis, this lead selection procedure can be used with other kinds of ECG based diagnostic tests by assisting cardiologists in minimizing interpretation errors.

(Abstract Control Number: 50)