On‐line size measurement of yeast aggregates using image analysis

Abstract

Alcohol fermentation productivity can be strongly improved using a flocculation-based yeast recycle. However, the efficiency of the biomass retention system depends strongly on the yeast particle size. Accordingly, the monitoring and control of yeast floc diameter are of primary importance. The on-line measurement of mean floc diameter has been achieved using on-line image analysis, based on the evaluation of image texture. The texture analysis method consisted in the building of a co-occurrence matrix from which the so-called "Energy parameter" was extracted. While image texture is usually used for classification purposes, it has been used here as a quantitative descriptor: a correlation has been found between this statistical image feature and off-line manual floc-size determinations. In the floc-size range investigated (X 0.5-4.3 mm), the evaluated mean diameter was in good agreement with the actual particle size, with a determination coefficient equal to 0.980. In contrast with manual measurements, slow and tedious, this method gave the value of the mean particle diameter in real-time, without sampling. This novel tool has been used to investigate the behavior of yeast aggregates as a function of fermentation conditions. While biomass concentration was kept constant, step increases of the feed rate led to a decrease of the mean floc diameter. Image analysis showed that the particle-size reduction could occur within a few minutes after modification of the medium dilution rate, demonstrating the disruptive effect of the CO(2) efflux. The kinetic of aggregate formation was dependent on the gas-phase composition. Instead of recycling fermentation gas, sparging the fermentor with nitrogen, to reduce dissolved CO(2) concentration, increased the rate of floc-size growth.

Contribution of image analysis to the description of enzymatic degradation kinetics for particulate food material

“…An alternative is to use a CCD video camera to visualise the particles and to extract size distribution by image analysis techniques. Several papers report the use of imaging to follow the evolution of the physical state in a reactor (Mas & Ghommidh, 2001;Sarker, Bertrand, Chtioui, & Popineau, 1998). Observing enzymatic degradations usually performed in stirred-tank reactor was a tricky challenge without sampling in the media.…”

“…Several methods have been proposed for extracting textural features from images (Haraclick, 1979) among which some have been proposed to characterise particle size. Co-occurrence features have proved efficient for correlation with the mean particle diameter and for quantifying changes in yeast floc size during continuous fermentation (Mas & Ghommidh, 2001). Constant grey level run length has been applied for the characterisation of bulk pea powders (Bertrand, Robert, Melcion, & Sire, 1991).…”

Flocculation in Saccharomyces Cerevisiae

“…On-line measurement of yeast fl occulation has been attempted by several authors ( Van Hamersveld et al , 1993 ;Mas and Ghommidh, 2001 ). Recently, a new method was proposed by Jibiki et al (2001) based on polymerase chain reaction (PCR) amplifi cation of the FLO5 gene.…”

Comparison of three coagulants by online turbidity monitoring

“…Thus, the average floc diameter measured via floc monitoring has been studied to index the coagulation efficiency. Floc monitoring techniques such as microscopy, microscopy combined with photography or image analysis, light scattering and photometric dispersion analysis (PDA) have been used to measure qualitative changes in floc size (Gregory, 1985(Gregory, , 2004Han et al, 2006;Huang, 2005;Mas and Ghommidh, 2001;Wang et al, 2005).…”