THE SKIN SENSITIVITY INDEX: PAVING THE WAY TO OBJECTIVE EVALUATION AND GRADING OF SKIN SENSITIVITY
427
Presented by: Juliette Rengot
INTRODUCTION
Sensitive skin is one of the most common disturbing skin conditions and highly impacts life quality. According to Symrise consumer and market insights, 56% of subjects declare having sensitive skin, a large and increasing number. A quarter of them considers it is a persistent condition [1]. As first perceptions of their skin sensitivity, consumers identify a skin dryness, an itching sensation, as well as the appearance of pimples and redness.
However, sensitive skin is still now solely based on consumers self-perception. There is still no analytical tool and approach to evaluate skin sensitivity. In this study, we present a method to analyze the skin in a multi-parametric way, to calculate linear skin sensitivity indices.
METHODS
An in vivo study was performed with 70 subjects self-declaring to have sensitive skin and 20 subjects self-declaring to have non-sensitive skin. Data were obtained by a self-assessment questionnaire, expert assessment of sensitivity status based on an interview with the subjects, analysis of hyperspectral images (melanin, hemoglobin, and oxygen saturation values, obtained with SpectraFace device) [2,3] in different areas of the face. Additionally, a sting test with lactic acid was performed.
To better understand skin sensitivity, we computed correlations between all parameters and the sensitivity status defined by experts. It enables us to select the most relevant information to construct a clinical skin sensitivity index. We used a linear regression of the responses to the questionnaires (sensitivity ranging from 0 to 10). To validate the obtained index, we compared it with the sensitivity status given by an expert (binary assessment of "sensitive status" versus "non-sensitive status").
To go further, we defined an instrumental sensitivity index based on hyperspectral images acquired with the SpectraFace. From hyperspectral data, both color and chromophore concentrations were computed and evaluated on the cheek, the cheekbone, and the nasolabial fold areas. A neural network was trained on these parameters to model the clinical sensitivity index defined previously.
RESULTS
Analysis of the lactic acid sting test results shows no correlation with the questionnaire-based skin sensitivity self-perception of subjects. Thus, its results were not considered to calculate the skin sensitivity indices.
The clinical sensitivity index achieved a 96,8% precision in classifying according to their status defined by experts (sensitive versus non-sensitive). The instrumental sensitivity index has a 90% correlation with the clinical sensitivity index and an 82% precision with the sensitivity status defined by experts. Both indices allowed having a continuous and fine sensitivity scale, either based on subjects’ perceptions or instrumental acquisitions.
DISCUSSION AND CONCLUSION
As expected, skin sensitivity is a multi-parametric condition. In this study, we showed that a “classical” sensitivity evaluation like the lactic acid test is not suitable to grade sensitivity. Instead, we propose an innovative tool to analyze skin sensitivity that allows calculating skin sensitivity indices from relevant information of a self-perception questionnaire and/or instrumental measurements. This innovative tool could support the assessment of skin sensitivity evolution under environmental changes, over time, or upon using a soothing treatment…
Further steps are necessary to improve the indices’ reliability and stability. These include working with a larger panel presenting uniform skin sensibility repartition. Indeed, the bigger the dataset, the better are the results of the neural network. Reinforcing the ground truth by asking several times the questionnaires should make our instrumental sensitivity index more trustworthy.
[1] Symrise CICS (Cosmetic Ingredients consumer study) database 2012 – 2018
[2] Model-based Skin Pigment Cartography by High-Resolution Hyperspectral Imaging. P. Seroul et. al. Journal of Imaging Science and technology 2016
[3] SpectraCam®: A new polarized hyperspectral imaging system for repeatable and reproducible in vivo skin quantification of melanin, total hemoglobin, and oxygen saturation. A. Nkengne et. al. Skin Res Technol. 2017
Sensitive skin is one of the most common disturbing skin conditions and highly impacts life quality. According to Symrise consumer and market insights, 56% of subjects declare having sensitive skin, a large and increasing number. A quarter of them considers it is a persistent condition [1]. As first perceptions of their skin sensitivity, consumers identify a skin dryness, an itching sensation, as well as the appearance of pimples and redness.
However, sensitive skin is still now solely based on consumers self-perception. There is still no analytical tool and approach to evaluate skin sensitivity. In this study, we present a method to analyze the skin in a multi-parametric way, to calculate linear skin sensitivity indices.
METHODS
An in vivo study was performed with 70 subjects self-declaring to have sensitive skin and 20 subjects self-declaring to have non-sensitive skin. Data were obtained by a self-assessment questionnaire, expert assessment of sensitivity status based on an interview with the subjects, analysis of hyperspectral images (melanin, hemoglobin, and oxygen saturation values, obtained with SpectraFace device) [2,3] in different areas of the face. Additionally, a sting test with lactic acid was performed.
To better understand skin sensitivity, we computed correlations between all parameters and the sensitivity status defined by experts. It enables us to select the most relevant information to construct a clinical skin sensitivity index. We used a linear regression of the responses to the questionnaires (sensitivity ranging from 0 to 10). To validate the obtained index, we compared it with the sensitivity status given by an expert (binary assessment of "sensitive status" versus "non-sensitive status").
To go further, we defined an instrumental sensitivity index based on hyperspectral images acquired with the SpectraFace. From hyperspectral data, both color and chromophore concentrations were computed and evaluated on the cheek, the cheekbone, and the nasolabial fold areas. A neural network was trained on these parameters to model the clinical sensitivity index defined previously.
RESULTS
Analysis of the lactic acid sting test results shows no correlation with the questionnaire-based skin sensitivity self-perception of subjects. Thus, its results were not considered to calculate the skin sensitivity indices.
The clinical sensitivity index achieved a 96,8% precision in classifying according to their status defined by experts (sensitive versus non-sensitive). The instrumental sensitivity index has a 90% correlation with the clinical sensitivity index and an 82% precision with the sensitivity status defined by experts. Both indices allowed having a continuous and fine sensitivity scale, either based on subjects’ perceptions or instrumental acquisitions.
DISCUSSION AND CONCLUSION
As expected, skin sensitivity is a multi-parametric condition. In this study, we showed that a “classical” sensitivity evaluation like the lactic acid test is not suitable to grade sensitivity. Instead, we propose an innovative tool to analyze skin sensitivity that allows calculating skin sensitivity indices from relevant information of a self-perception questionnaire and/or instrumental measurements. This innovative tool could support the assessment of skin sensitivity evolution under environmental changes, over time, or upon using a soothing treatment…
Further steps are necessary to improve the indices’ reliability and stability. These include working with a larger panel presenting uniform skin sensibility repartition. Indeed, the bigger the dataset, the better are the results of the neural network. Reinforcing the ground truth by asking several times the questionnaires should make our instrumental sensitivity index more trustworthy.
[1] Symrise CICS (Cosmetic Ingredients consumer study) database 2012 – 2018
[2] Model-based Skin Pigment Cartography by High-Resolution Hyperspectral Imaging. P. Seroul et. al. Journal of Imaging Science and technology 2016
[3] SpectraCam®: A new polarized hyperspectral imaging system for repeatable and reproducible in vivo skin quantification of melanin, total hemoglobin, and oxygen saturation. A. Nkengne et. al. Skin Res Technol. 2017