The P1 latency of single-trial ERPs estimated by two peak-picking strategies

Reliable latency estimation of a given component is essential for single-trial analysis of event-related potentials (ERPs). Peak-picking is a straightforward method for latency determination in single trials. The peak-picking method has so far been carried out according to the maximal-peak strategy, which looks for the largest deflection in a low-pass filtered epoch within a predefined time window. We introduced another peak-picking criterion and named it as the closest-peak strategy in the present study. According to the strategy, the peak of an ERP component in single trials is defined as the peak timely closest to the average-ERP peak. The main results derived from visual stimulus ERP data at POz showed that: 1) the single-trial P1 latency estimated by the closest-peak strategy was timely concentrated around the P1 peak in average ERP while the P1 peak extracted by the maximal-peak strategy distributed relatively scattered on the time axis; 2) the P1 amplitude in average ERP acquired by the maximal-peak strategy was larger than that acquired by the closest-peak strategy at 5% significance level. Although the conclusion is premature, this comparative study suggested that the closest-peak strategy may be more suitable and reliable for peak-picking method.