Automated Image-Based Color Analysis as an Accessible and Widely Applicable PAT Tool
Simon Smolders, Huibo Sheng, Matthew P. Mower, Aditi Potdar, Jan Dijkmans
Abstract
Pharmaceutical R&D is increasingly employing data-driven approaches with specialized process analytical technology (PAT) tools to develop processes efficiently and reliably. However, some unit operations are still developed based on human observations, which lack quantification and objectivity and require personnel attending the experiment. In this article, we demonstrate the ease of using cameras to provide data in nearly every unit operation. While applications of basic to advanced image analysis algorithms have been published, herein we use a simple color analysis which excels in its broad applicability without the need for model training. There is currently little literature on color-based analysis, and the few published papers target only one specific use. The color analysis was able to provide valuable information on reaction progress, color removal, level monitoring, settling times, solubility determination, nucleation detection, filtration curves, and color of isolated materials without any required tuning of the analysis. The data from the color analysis often matched or outperformed the state-of-the-art PAT tools.