Figure 3.

Comparison of FLIM bi-exponential fitting and Pseudo-FLIM methods for melanin detection. FLIM analysis – 2PEF intensity and τ₁ short-fluorescence lifetime images of (a) synthetic melanin and (b) normal human skin melanocytes and keratinocytes NHMK coculture. The FLIM τ₁ – based melanin mask is obtained by application of a threshold to keep the pixels with τ₁ values below 80 ps. (c) 2PEF intensity decays of synthetic melanin, melanin and non-melanin pixels within the NHMK coculture (averaged 2PEF intensity within the green and respectively red regions of interest). (d) Normalized histograms of the pixel frequency for ${\tau }_1$ FLIM images of synthetic melanin (a) and NHMK coculture (b). The blue rectangle depicts the histogram regions selected by the application of a τ₁ < 80 ps threshold. (e) Pseudo-FLIM analysis of NHMK coculture, involving (1) a temporal binning of the 2PEF decay into a reduced number of time channels with 2 ns integration time per channel, followed by (2) an estimation of the slope of the decay after linear regression on the first 3 t-channels. (f) Example of resulting binned 2PEF decay of a melanin pixel (red arrow) and a non-melanin pixel (yellow arrow) selected based on their τ₁ value. (g) After a natural logarithm transformation, a linear regression of the first 3-time channels is performed to calculate the slope of the decay. The faster the decay, the higher the slope, as observed for these two melanin and non-melanin pixels. The image of the Pseudo-FLIM slope parameter in (e) is further processed for melanin detection to keep the high slopes values above 70 arb. u. and create a Pseudo-FLIM melanin mask.