All-numerical noise filtering of fluorescence signals for achieving ultra-low limit of detection in biomedical applications.

We present an all-numerical method for post-processing of the fluorescent signal as obtained from labeled molecules by capillary electrophoresis (CE) in an optofluidic chip, on the basis of data filtering in the Fourier domain. It is shown that the method outperforms the well-known lock-in amplification during experiments in the reduction of noise by a factor of (square root)2. The method is illustrated using experimental data obtained during CE separation of molecules from a commercial DNA ladder with 17 fluorescently labeled molecules having different base-pair sizes. An improvement in signal-to-noise ratio by a factor of ∼10 is achieved, resulting in a record-low limit of detection of 210 fM.