About Jean Christophe PITTET
Pittet is a seasoned R&D consultant specializing in skin science and dermocosmetic product evaluation. With over 25 years of experience, he founded and led ORION-Concept, a scientific consultancy focused on skin research and cosmetic innovation. Previously, he served as Scientific Director and Deputy General Manager at leading institutes, including the Institut d’Expertise Clinique and Spincontrol, where he oversaw international teams and developed cutting-edge methods for assessing skin tolerance and product efficacy. Holding a PhD in Life and Health Sciences, he has also taught at the university level and trained professionals in skin metrology and imaging. Actively involved in scientific communities, he co-founded the Dermaloire Scientific Interest Group and serves as a board member of the Francophone Society of Cutaneous Engineering and Imaging. A pioneer in biomedical imaging and bioengineering, Jean-Christophe bridges scientific expertise with education to advance dermatological research.
The theme of the radiance of the complexion is the subject of our ZOOM#34 dossier, and we would like to know how, in your opinion, the in vivo evaluation of this claim has evolved in recent years
Anyone can easily judge the radiance of a complexion – “healthy glow”, “radiant complexion”, “fresh complexion”… etc., but precisely defining its components remains a delicate undertaking. Multiples, often depending on the phototype, vary according to the individual interpretation. To objectify this evaluation, a method has been defined in collaboration with Spincontrol (now Eurofins) and L’Oréal (Visual evaluation in vivo of “complexion radiance” using the C.L.B.T. sensory methodology, Musnier et al., 2004). This is based on a visual scale of hues and unstructured sensory descriptors (skin texture, homogeneity, transparency, areas of light reflection). This external visual reference allowed for the calibration of the evaluators, reducing the subjectivity. A specific adaptation for Asian skin was made in 2012.
On this basis, an image analysis method was developed, taking up the clinical criteria: distribution of shades according to the areas of the face, skin texture, shine (specular vs. diffuse), transparency and colorimetric homogeneity. Two fundamental dimensions emerged: Clarity, linked to the intensity of the hues (increase in L* in the CIELab space), and Luminosity, dependent on the reflection of light by the skin.
However, one major variable is still little considered: facial expression. A complexion, even a radiant one, can appear dull if the face expresses anxiety or sadness. Artificial intelligence could offer a new way of integrating these affective parameters, which are still largely underexplored.
What differences would you make between the radiance of the complexion and the evenness of the complexion?
As always, we would like to boil down a complex problem to simple things. Homogeneity is clearly an important (essential?) element in the problem of evaluating the radiance of the complexion but cannot define it on its own… A phototype 1, red and scabbed with freckles, could not have a beautiful shine under the pretext of “heterogeneity”? A bit restrictive, isn’t it?
This uniformity is even a limitation in some cases depending on the scale of observation: pinkness on the cheekbones and protruding regions is often a sign of brightness and “freshness” although creating “color heterogeneities”.
Which in vivo methods do you think are the most suitable: scoring, photos, biometrology, self-assessment, etc.
As mentioned earlier, the published CLBT method provides a solid basis for approaching the evaluation of “complexion radiance”. The result of a long process of defining the essential criteria is based on clinical evaluations carried out by trained judges, capable of perceiving the various parameters with finesse. It allows a certain standardization of the approach. In addition, calibrated photography remains a so-called “objective” method, provided that several elements are integrated: color distribution according to the areas of the face, clarity of shades, modalities of brilliance (“luminosity”), skin texture, surface homogeneity, among others. Other more secondary parameters can also enrich the analysis.
In biometrology, the difficulty persists: no single device can measure all the aspects related to the radiance of the complexion. The devices designated as “radiance-meters” are not enough to capture all its dimensions. One potential solution lies in the combination of several instruments, each targeting a key parameter.
Finally, self-evaluation remains essential. It is an essential check to verify the user’s acceptance of the result. This principle goes beyond the sole framework of the radiance of the complexion: any cosmetic effect, to be relevant, must be perceived – objectively or subjectively – by the consumer. Without this membership, there is no success.
In the end, an integrative clinical approach, capable of encompassing all the parameters defining radiance, remains not only relevant, but essential. Instrumentalist’s word!
In your opinion, what are the essential elements to consider when designing them in vivo study?
My answer will be generic. Depending on the characteristics that we are trying to evaluate in their evolution under the effect of treatment/care, the sample must be selected with particular care. Too much diversity in the panel (phototype, skin type, ethnicities, etc.), generally small (25 to 40 on average), will imply a great diversity of response and therefore a great dispersion in these evolutions… Statistical validation will be lost even if the product shows real activity. The argument is often to be “representative” of a population. Is this reasonable for 30 volunteers? The homogeneity of the panel (excluding age) is to be preferred in all cases.
For the specific case of radiance, it is necessary to first define what needs to be studied: A “healthy glow” effect, an improved “radiance”, a “homogenization” of the complexion… etc. This will depend on the claims as well as on the formulation and the objective activity of the product on certain skin features. The selection of subjects will be made based on these criteria, which we seek to change under the effect of the product. An obvious choice for some, but unfortunately, I’ve seen too many anti-wrinkle studies, for example, on subjects… without wrinkles or too little!
Finally, how do you think AI can bring benefits to this evaluation through the automatic evaluation of photos or algorithms?
Artificial intelligence (AI) represents a major methodological lever in the evaluation of complex aesthetic parameters such as the radiance of the complexion, characterized by its intrinsically multi-parametric nature. This phenotype results from the integration of cutaneous variables (chromatic distribution, microrelief, specular/diffuse shine, pigmentary homogeneity) and contextual parameters such as facial expressions, which are often ignored.
The main interest of AI lies in its ability to model nonlinear interactions between these multiple dimensions. However, most current implementations remain purely predictive in nature, without shedding light on the underlying mechanisms. A more explanatory approach, based on the analysis of weights, correlations and interaction effects between characteristics, would be necessary for a detailed understanding of the determinants of perceived brightness.
The massive training on generalist databases (representing a very wide inter-individual diversity) also raises questions. A more relevant strategy in a cosmetological context would be to develop specialized AIs, restricted to targeted phenotypes (brightness, clarity, uniformity), and then integrate these expert modules into a composite or multi-agent architecture.
In addition, considering micro-expressions, as modulators of the perception of radiance, opens a new path. The integration of emotional facial recognition networks into AI models would make it possible to contextualize predictions, considering the cognitive impact of affective states on the perception of skin quality.
Thus, AI, well mastered, could become a powerful aesthetic quantification tool, combining instrumental robustness and perceptual sensitivity.
Let’s trust in the future!
