Emulsion Quality Evaluation Using Automated Image Analysis

EasyChair Preprint 8762

Abstract

An emulsion represents a mixture of minute droplets of one liquid in another, which are normally non-mixable and non-blendable. The evaluation of emulsion quality is typically based on human assessment of microscopic images. This can lead to subjectivity in the interpretation as to whether the product meets requirements or not. An alternative approach is to automate the analysis of micrographs using image processing software that measures the emulsion quality based on the droplet characteristics.
In this study, an automated image analysis of optical micrographs has been applied to detect the individual droplets and agglomerates and their characteristics in oil-in-water emulsions. A statistical examination of the characteristics obtained through image analysis is performed to evaluate the requirements of a desirable sample. The characteristics investigated in this study are droplet size and dispersion. Droplet size is measured in terms of area and Feret diameter and droplet dispersion is studied based on agglomeration.
A set of samples with desirable and undesirable properties were used to perform this investigative study. Absence of larger agglomerates of size greater than 150 μm2, occasional presence of agglomerates (of size between 80 and 150 μm2) and a droplet size distribution with more than 60% of the droplets having a size less than 3 μm2 are identified as the desirable emulsion properties. The study proved that automated image analysis has the potential to assess the quality of emulsions in a more sophisticated way compared to human analysis. Therefore, the method can be recommended as a tool for in-situ process monitoring and real time quality control in emulsion preparation.

Keyphrases: Droplet characteristics, Emulsion, automated image analysis

@booklet{EasyChair:8762,
  author    = {Saritha Unnikrishnan and John Donovan and David Tormey and Russell Macpherson},
  title     = {Emulsion Quality Evaluation Using Automated Image Analysis},
  howpublished = {EasyChair Preprint 8762},
  year      = {EasyChair, 2022}}