Quantitative Estimation of Long-living Fluorescent Molecules from Temporal Fluorescence Intensity Data Corrupted by Nonzero-mean Noise

We present a new quantitative method of estimation of fluorescent molecule numbers from time-lapse, single-cell, fluorescence microscopy data. Its main aim is to eradicate backward propagation of noise, which is present in previous methods. The method is first validated using Monte Carlo simulations. These tests show that when the time-lapse data are corrupted with negative noise, the method obtains significantly more precise results than current techniques. The applicability of the method is demonstrated on novel time-lapse, single-cell measurements of fluorescently tagged ribonucleic acid (RNA) molecules. Interestingly, we find that the intervals inferred by the new method have the same mean but reduced variability when compared to the previously existing method, which, in accordance to human observers, is a more accurate estimation.