Sizing Lamellar Cosmetic Emulsions: Deep-Learning Image Segmentation and Laser Diffraction as Complementary Techniques

Droplet size distribution governs the stability and sensory properties of cosmetic emulsions. Many cosmetic emulsions are lamellar. Each droplet carries a birefringent shell and is a core-shell object. Laser diffraction is the usual sizing technique. Its Mie inversion requires a single refractive index. This index is poorly defined for a droplet with a lamellar shell. Optical microscopy avoids this assumption. It measures each droplet directly on the image, whatever its refractive index. Combined with image segmentation, it offers an independent sizing route. We compared laser diffraction with a classical watershed and a deep-learning model (Cellpose) on 13 lamellar emulsions imaged at ×10, ×20, and ×40. Diameters were calibrated from the instrument metadata and reweighted to volume. Cellpose and laser diffraction give the same central size. Their dv50 differ by less than 1.5 µm on all 13 emulsions at ×10. Paired tests show no significant dv50 difference at ×10 and ×40 (adjusted p ≥ 0.13). The watershed reaches this agreement only on the coarsest emulsions. It overestimates the dv50 of the finer ones by 15 to 35% and fails at ×40. The two techniques differ only at the ends of the distribution. At the upper end, laser diffraction gives a higher dv90 than microscopy. A non-spherical laser model does not remove this gap. Possible causes are the smoothing of the laser inversion and the evolution of the sample under dilution. At the lower end, microscopy gives a higher dv10 because it does not detect droplets below its optical resolution. The same slides also reveal the lamellar shell under crossed polarizers. Deep-learning microscopy is thus a quantitative sizing method for the central size of lamellar emulsions. The two techniques are complementary for the characterization of lamellar emulsions.

Authors

Institutions

Publication Details

Journal
Cosmetics
Published
2026-11-19
DOI
https://doi.org/10.3390/cosmetics13060323
Primary Topic
Surfactants and Colloidal Systems
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Sizing Lamellar Cosmetic Emulsions: Deep-Learning Image Segmentation and Laser Diffraction as Complementary Techniques

Philippe Piccerelle, Christophe Sauzet, Magalie Claeys‐Bruno, Raphaël Coatmeur et al.
Cosmetics
Surfactants and Colloidal Systems
article

Sizing Lamellar Cosmetic Emulsions: Deep-Learning Image Segmentation and Laser Diffraction as Complementary Techniques

Philippe Piccerelle, Christophe Sauzet, Magalie Claeys‐Bruno, Raphaël Coatmeur, Axel Viton, Béatrice Anthouard
article en

Abstract

Droplet size distribution governs the stability and sensory properties of cosmetic emulsions. Many cosmetic emulsions are lamellar. Each droplet carries a birefringent shell and is a core-shell object. Laser diffraction is the usual sizing technique. Its Mie inversion requires a single refractive index. This index is poorly defined for a droplet with a lamellar shell. Optical microscopy avoids this assumption. It measures each droplet directly on the image, whatever its refractive index. Combined with image segmentation, it offers an independent sizing route. We compared laser diffraction with a classical watershed and a deep-learning model (Cellpose) on 13 lamellar emulsions imaged at ×10, ×20, and ×40. Diameters were calibrated from the instrument metadata and reweighted to volume. Cellpose and laser diffraction give the same central size. Their dv50 differ by less than 1.5 µm on all 13 emulsions at ×10. Paired tests show no significant dv50 difference at ×10 and ×40 (adjusted p ≥ 0.13). The watershed reaches this agreement only on the coarsest emulsions. It overestimates the dv50 of the finer ones by 15 to 35% and fails at ×40. The two techniques differ only at the ends of the distribution. At the upper end, laser diffraction gives a higher dv90 than microscopy. A non-spherical laser model does not remove this gap. Possible causes are the smoothing of the laser inversion and the evolution of the sample under dilution. At the lower end, microscopy gives a higher dv10 because it does not detect droplets below its optical resolution. The same slides also reveal the lamellar shell under crossed polarizers. Deep-learning microscopy is thus a quantitative sizing method for the central size of lamellar emulsions. The two techniques are complementary for the characterization of lamellar emulsions.

CosmeticsVol. 13(6)
Centre National de la Recherche Scientifique (FR), Aix-Marseille Université (FR), Institut Méditerranéen de Biodiversité et d'Ecologie Marine et Continentale (FR), Institut de Recherche pour le Développement (FR)
Surfactants and Colloidal Systems
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon