The aim of this work was to develop approaches to utilize whole particle distributions for both particle size and particle shape parameters to map the full range of particle properties in a curated dataset. It is hoped that such an approach may enable a more complete understanding of the particle landscape as a step towards improving the link between particle properties and processing behaviour. A 1-dimensional principal component analysis (PCA) approach was applied to create a 'morphological distribution landscape'. A dataset of imaged APIs, intermediates and excipients encompassing particle size, particle shape (elongation, length and width) and distribution shape was curated between 2008 and 2022. The curated dataset encompassed over 200 different materials, which included over 150 different APIs, and approximately 3500 unique samples. For the purposes of the current work, only API samples were included. The morphological landscape enables differentiation of materials of equivalent size but varying shape and vice versa. It is hoped that this type of approach can be utilised to better understand the influence of particle properties on pharmaceutical processing behaviour and thereby enable scientists to leverage historical knowledge to highlight and mitigate risks associated to materials of similar morphological nature.
Keywords: Image analysis; Machine learning; Materials science; Morphology; Particle size; Physical characterization.
Copyright © 2023 Elsevier B.V. All rights reserved.