Skip to main content
. 2024 Sep 24;29(9):093509. doi: 10.1117/1.JBO.29.9.093509

Fig. 2.

Fig. 2

Two strategies for collecting the training dataset. Strategy (a) in which we train a network on synthetic attenuation-concentration pairs generated from the modified Beer–Lambert law. Strategy (b) in which the training is performed on pairs of real attenuation and concentrations obtained through the least-squares fit to the corresponding real spectra.