Background: Skin is the first line of defense against harmful external environmental factors. Skin flora living on the skin surface impact skin health and skin disease. Bacteria, form part of the unique and complex skin micro-ecological system. For example, Propionibacterium acnes (P. acnes) is a member of the anaerobic organisms and is involved in the induction of skin acne. It produces porphyrins that absorb ultraviolet light and emit red fluorescence in response. As a result, fluorescence surveillance of the skin can be important in both the diagnosis of skin acne and the evaluation of therapeutic effects. Many different measurement methods for single skin biophysical properties have been reported.. This study focused on the age-dependent changes in porphyrins for normal skin, and developed a novel algorithm to evaluate porphyrins using the fluorescence images by image processing quantitatively.
Materials and methods: An extraction algorithm was proposed for the segmentation of porphyrin fluorescence images in OpenCV. The algorithm consisted primarily of preprocessing, conversion from RGB color space to HSV color space, and classification of fluorescence. There are 3595 healthy Japanese aged 16-85 years enrolled in the study and fluorescence images were acquired from their cheek sites under 375 nm UV-LED excitation. Age-related fluorescence variation was conducted applying the algorithm implemented.
Results: A new extraction algorithm has been proposed with fluorescence image input and three indexes output, including the number of fluorescence, area of fluorescence, and mean intensity of fluorescence. Proposed algorithm was verified by three parameters, the accuracy, sensitivity, and precision, which refer to the ability of algorithm to detect the number of fluorescence correctly and repeatedly. The verification results were 71%, 72%, and 88% respectively, taking a validly fundamental step for skin health record and analysis. Furthermore, large-scale fluorescence image segmentation results revealed that similar trends were coming out for all three indexes in cheek as people get older. All the fluorescence number, area and mean intensity arrived at the highest at 30 years old and fell off since then.
Conclusion: The number, area, and fluorescence intensity of porphyrins can be extracted well from fluorescence images with the proposed algorithm in the study, which has the potential to aid in thediagnosis of skin acne and predict skin conditions as an assisted tool. It is implicated that fluorescence status is influenced by age, which rises to the peak around 30 years old for normal cheek's skin.
Keywords: Age; HSV color model; Image processing; P. acnes; Porphyrin; Portable device.
Copyright © 2021 The Author(s). Published by Elsevier B.V. All rights reserved.