Optimal algorithm for single-molecule identification with time-correlated single-photon counting

A pattern-matching procedure for identifying single molecules according to the temporal decay characteristics of the fluorescence is presented and applied to measured single-molecule data of three different dyes. Based on the maximum likelihood principle, this procedure is the best possible identification algorithm if the fluorescence time decays of the different molecular species are exactly known. The presented algorithm is numerically simple and fast and, thus, especially suitable to possible online applications. Exact theoretical results for;he misidentification errors of the algorithm are derived and compared with the errors determined from the experimental data.