Open AccessQualification of a new and precise automatic tool for the assessment of hair diameters in phototrichograms
Author(s) -
Scheede Stefan,
Herpens Andreas,
Burmeister Frank,
Oltrogge Bernhard,
Saenger K.,
SchmidtRose Thomas,
Schreiner Volker,
Wenck Horst,
Knieps Torsten,
Berlage Thomas
Publication year - 2011
Publication title -
skin research and technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.521
H-Index - 69
eISSN - 1600-0846
pISSN - 0909-752X
DOI - 10.1111/j.1600-0846.2010.00482.x
Subject(s) - modular design , computer science , software , artificial intelligence , pixel , computer vision , focus (optics) , annotation , gold standard (test) , biomedical engineering , mathematics , statistics , medicine , optics , physics , programming language , operating system
Background/purpose: To automatically assess hair growth during cosmetic trials, incorporating parameters such as anagen‐to‐telogen rate, growth rate, and especially hair diameter. Methods: We designed and qualified a new and automatic phototrichogram system based on a high‐resolution DSLR camera system (theoretical resolution of 2.557 μm/pixel) and modular macrolens system with fixed focus, combined with a trainable pattern recognition software for automated analysis. Results: We improved the standard routine for dermatological phototrichogram technique to overcome inaccuracy in thickness measurements due to hair swelling by using an alternative immersion fluid, and increased the effective resolution for hair size and thickness measurement to <4 μm. After having qualified manual measurements as gold standard for the determination of hair diameters, we established a new trainable automatic picture analysis software able to locate and measure individual hairs in length and thickness even in picture series taken from the same skin area at different time points. Comparisons between manual and automatic measurements of the same hairs showed a >90% correlation, and by comparing the automatic results with manual measurements of the same images without individual hair annotation, we could find a correlation of at least 80%. Conclusion: According to the results and findings generated in this qualification study, we have a reliable tool now that enables us to test cosmetic products for hair treatment in a highly automated way with a sufficient degree of precision and accuracy to detect even small changes in hair diameter during cosmetic trials.