How to get reliable QT measurements when your ECG is noisy: example for 24-hour rabbit ECG
15 Sep 2014


Analysis of ECG signals recorded from small animals is often tricky due to low signal to noise ratio.

An ECG signal was acquired by telemetry on a rabbit and the automated analyzer ECG30g as well as the QT correction module QTC10a were used to perform QT analysis (the configuration is shown in Figure 1). The analysis results without QT correction are shown on the 2 first rows of Figure 3. Placement of T ends in certain noisy areas appear to be low in particular.


In order to tentatively improve the analysis, a new configuration was used with the pattern averaging filter PTS11a, as shown in Figure 2. PTS11a calculates a point-to-point average over a number of signal cycles. The QRS complex detector QRS10a was used to trigger PTS11a.

Quantification of the analysis from raw compared to filtered signal is shown in table 1. Coverage of T end placement in particular was increased from 39.6 % to 97.9%.

In conclusion, the pattern averaging filter PTS11a was used to successfully improve the coverage and stability of QT analysis on noisy rabbit ECG telemetry signal.


Figure 1: Configuration for raw ECG analysis.


Figure 2: Configuration for filtered ECG analysis.


Figure 3: ECG30g analysis of raw and filtered rabbit ECG signals showing the whole 7h acquisition (A), a first zoom on the orange zone (B) and then a stronger zoom (C). First row correspond to raw ECG, 2nd row to QT duration derived from raw ECG analysis, 3rd row is filtered ECG and lowest row is QT duration derived from filtered ECG analysis.


Table 1. Quantification of the QT analysis on raw and filtered rabbit noisy ECG. Percentages are calculated by ratio with R peaks detected by QRS10a.



Filtered ECG


Number of detected R peaks

49177 (55.1%)

88951 (99.7%)


Number of detected T ends

35356 (39.6%)

87359 (97.9%)


QT interval average




QT interval Std dev