Electrical signals measured by microelectrode arrays (MEAs) allow non-invasive long-term monitoring of cardiomyocyte cultures. However, variability of local responses between electrodes and noise makes difficult the accurate estimation of the field potential duration (FPD). False peaks are often detected and falsify the FDP estimated value, which finally requires for the user to perform some manual corrections for each well. This drawback not only slows down the analysis but also introduces arbitrary estimations and inter-individual errors. The main objective is to develop a new estimation method of FPD taking advantages of all the information contained in the signals measured by all the electrodes of the same well. The motivations are to automatically compute FPD and to improve the accuracy and robustness of the estimation compared to existing techniques already implemented in the available technologies.