Abstract :
The restitution of the action potential duration (APDR) is an important physiological property of the cardiac tissue and it is related to the development of fatal arrhythmia. Accurate measures of the slope of the APDR curve are crucial for a correct interpretation of the restitution properties. The two most commonly used methodologies for the estimation of the APDR slope are based on linear and exponential fitting. In this study, an accurate assessment and comparison of these methodologies is performed for the first time. Realistic APDR curves were simulated based on an analytical model of the APDR slope that was informed with data derived from unipolar electrograms recorded in structurally normal human ventricles. The impact of noise level was evaluated. When the activation and repolarization times for both the basic cycle length beat and the premature beat were measured with an error <; 5 ms, the linear method provided more accurate estimates, being its mean absolute estimation error significantly lower (11:3% - 25:3% vs 37:0% - 38:2%) and its linear correlation with reference values significantly higher (0:97 - 0:98 vs 0:65 - 0:70) than for the exponential method. These relatively high estimation errors and high correlations suggest that restitution properties should be considered in relative terms, and discourage the use of fixed threshold values, such as 1.00, for the interpretation of the slope estimates. In conclusion, this study suggests to use a piece-wise linear fitting scheme for the estimation of the APDR curve, provided that all measured with an error lower than 5 ms.
