Microalgae Monitoring in Microscale Photobioreactors via Multivariate Image Analysis

Abstract

Microscale photobioreactors for microalgae growth represent an interesting technology for fast data production and biomass characterization; however, the small scale poses severe monitoring challenges, as traditional methods cannot be used. Non-invasive techniques are therefore needed to quantify biomass concentration and other culture properties, for example, pigment composition. To this purpose, a soft sensing approach based on multivariate image regression is proposed to exploit RGB images and/or PAM-imaging chlorophyll fluorescence. Different PLS (Partial Least Squares) regression models are used to estimate: (a) biomass concentration from the features extracted by RGB indices and/or PAM-imaging chlorophyll fluorescence measurements; and (b) Chlorophyll a content per cell from the features extracted by RGB indices and biomass concentration measurements. Every single model is aimed at characterizing the microalgae culture at different light intensities during batch growth. Results show that the proposed monitoring approach is as accurate as traditional measurement approaches and may represent a promising methodology for fast and inexpensive monitoring of microscale photobioreactors.