Geometrical method for interpolating S-peaks from cow ECG using a microcontroller

• Simple algorithm to improve the quality of cow ECG data using interpolation.
• The algorithm is fast enough to run in real-time on a microcontroller
• Useful for implantable devices to analyse beat-to-beat interval with high accuracy.

This paper describes a new and computationally efficient adaptive geometrical method for improving the resolution of electrocardiogram (ECG) S-peak detection. The target device is an ECG implant for dairy cows which provides limited calculation power. The method utilises only arithmetic operations and is adaptive to individual ECG curves and electrode installation effects. The method tracks the average heights of the sample points preceding and following the S-wave peak. Comparing these heights to the long-time average values provides an estimate of the offset of the peak sample from the actual peak moment. On average, the prediction of spline interpolation on the original 512 Hz turning point is 0.043 ms more accurate than our geometrical algorithm. Both methods result in less than a 0.7 ms difference from the original data. However, the proposed method is significantly more computationally efficient than cubic spline interpolation.