
==== Front
Skin Res Technol
Skin Res Technol
10.1111/(ISSN)1600-0846
SRT
Skin Research and Technology
0909-752X
1600-0846
John Wiley and Sons Inc. Hoboken

10.1111/srt.13643
SRT13643
Original Article
Original Articles
Comparison of facial skin ageing in healthy Asian and Caucasian females quantified by in vivo line‐field confocal optical coherence tomography 3D imaging
ALI et al.
Ali Assi 1
Colombe Lopez 2
Mélanie Pedrazzani 2
Agnes Pignol‐Lavoix 3
Meryem Nili 3
Samuel Ralambondrainy 1
Guénolé Grignon 1
Jean‐Hubert Cauchard 1
Rodolphe Korichi 1
Franck Bonnier https://orcid.org/0000-0003-3658-4792
1 fbonnier@research.lvmh-pc.com

1 LVMH Recherche, Saint Jean de Braye Paris France
2 DAMAE Medical Paris France
3 COMPLIFE Lyon France
* Correspondence
Bonnier Franck, 185 avenue de Verdun, 45804 Saint Jean de Braye, France.
Email: fbonnier@research.lvmh-pc.com

02 9 2024
9 2024
30 9 10.1111/srt.v30.9 e1364302 2 2024
19 2 2024
© 2024 The Authors. Skin Research and Technology published by John Wiley & Sons Ltd.
https://creativecommons.org/licenses/by-nc-nd/4.0/ This is an open access article under the terms of the http://creativecommons.org/licenses/by-nc-nd/4.0/ License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.

Abstract

Background

Quantitative biomarkers of facial skin aging were investigated in 109 healthy Asian female volunteers, aged 20 to 70 years.

Materials and Methods

In vivo 3D Line‐field Confocal Optical Coherence Tomography (LC‐OCT) imaging, enhanced by Artificial Intelligence (AI)‐based quantification algorithms, was utilized to compute various metrics, including stratum corneum thickness (SC), viable epidermal (VE) thickness, and Dermal‐Epidermal Junction (DEJ) undulation along with cellular metrics for the temple, cheekbone, and mandible.

Results

Comparison with data from a cohort of healthy Caucasian volunteers revealed similarities in the variations of stratum corneum and viable epidermis layers, as well as cellular shape and size with age in both ethnic groups. However, specific findings emerged, such as larger, more heterogeneous nuclei in both layers, demonstrated by an increase in nuclei volume and their standard deviation, and increased network atypia, all showing significant age‐related variations. Caucasian females exhibited a flatter and more homogeneous epidermis, evidenced by a decreased standard deviation of the number of layers, and a less dense cellular network with fewer cells per layer, indicated by a decrease in cell surface density.

Conclusion

Ethnicity‐wise comparisons highlighted distinct biological features specific to each population. Asian individuals showed significantly higher DEJ undulation, higher compactness, and lower cell network atypia compared to their Caucasian counterparts across age groups. Differences in stratum corneum and viable epidermal thickness on the cheekbone were also significant. LC‐OCT 3D imaging provides valuable insights into the aging process in different populations and underscores inherent biological differences between Caucasian and Asian female volunteers.

artificial intelligence
Asian female
Caucasian female
facial ageing
healthy skin
line‐field confocal optical coherence tomography
non‐invasive 3D imaging
source-schema-version-number2.0
cover-dateSeptember 2024
details-of-publishers-convertorConverter:WILEY_ML3GV2_TO_JATSPMC version:6.4.8 mode:remove_FC converted:02.09.2024
Ali A , Colombe L , Mélanie P , et al. Comparison of facial skin ageing in healthy Asian and Caucasian females quantified by in vivo line‐field confocal optical coherence tomography 3D imaging. Skin Res Technol. 2024;30 :e13643. 10.1111/srt.13643
==== Body
pmc1 INTRODUCTION

Measuring skin aging biomarkers in vivo provides valuable insights into the physiological changes associated with both intrinsic (chronological aging) and extrinsic factors (such as photoaging, environmental influences, and lifestyle choices). Various methods have been developed to enhance our understanding of the physiological mechanisms of aging and to support efficacy testing of topical cosmetic products. While clinical assessments by dermatologists allow the evaluation of macroscopic signs of aging through visual observation or tactile examination of different facial areas, instrumental assessment through biometrics and biophysical measurements, including skin hydration, Transepidermal Water Loss (TEWL), elasticity, and firmness, can be performed using established tools like the corneometer, tewameter and cutometer. In addition, high‐resolution photography and dermoscopy enable the visualization and quantification of various visible, surface‐level signs of skin aging, including wrinkles, fine lines, changes in skin texture, alterations in pigmentation, dullness and the presence of actinic keratosis. 1 , 2

In recent decades, substantial efforts have been dedicated to the development of intracutaneous imaging techniques enabling the visualisation of skin structures at deeper levels, particularly within the epidermal and dermal layers. Pioneering studies have utilized Optical Coherence Tomography (OCT) and High‐Frequency Ultrasound (HFUS) to non‐invasively examine (and quantify) architectural changes associated with skin aging in various populations. 3 , 4 Imaging techniques are based on distinct principles and modalities that determine their performance in terms of spatial and axial resolutions, as well as the depth of analysis. Consequently, OCT and HFUS are most prominently employed for assessing epidermal thickness, the integrity of the Dermal‐Epidermal Junction (DEJ), dermal thickness, and changes in dermal density associated with the aging process. They offer valuable insights into collagen and elastin content, density, and arrangement. 5 , 6 , 7 Notably, they can quantify parameters such as collagen fibre density, spacing, and organization, facilitating the measurement of age‐related alterations. 8 , 9 The Sub‐Epidermal Low Echogenic Band (SELEB) is an ultrasonographic feature associated with skin aging, and it is considered a normal age‐related change, representing a gradual reduction in the density of collagen fibres within the papillary dermis. 6

The introduction of in vivo microscopy, including Reflectance Confocal Microscopy (RCM), Multiphoton Microscopy (MPM), and Line‐field Confocal Optical Coherence Tomography (LC‐OCT), has provided the means for detailed examinations of microstructures and cellular patterns with spatial resolutions typically down to a few microns. 10 , 11 , 12 Their micrometric precision makes them well‐suited for the study of age‐related variations in stratum corneum and epidermal thickness, the structural organization of the DEJ, and the density and morphology of skin cells. The applications of the imaging methods now extend to the visualization of pigmentation through the quantification of melanin distribution but also to the observation of blood vessels to assess the microcirculation, as reported in the literature. 13

For instance, LC‐OCT allows for a depth of analysis reaching as far as 500 µm into the superficial dermis of human skin, while simultaneously achieving an isotropic spatial resolution of approximately 1 µm. 14 , 15 Its unique capacity for vertical (B‐scan) and en face (horizontal or C‐scan) section imaging enables rapid and comprehensive 3D in vivo skin examinations. 16 , 17 Beyond its utility in diagnosing skin lesions such as carcinomas, 14 , 18 lentigo maligna, 19 and actinic keratosis, 20 LC‐OCT has also demonstrated promise when coupled with Artificial Intelligence (AI) for investigating the intricacies of healthy skin and age‐related modifications. 21 , 22 Notably, segmentation algorithms have emerged as promising tools for deriving a range of 3D quantitative parameters. These parameters encompass measurements of skin layer thicknesses, including the stratum corneum (SC) and Viable Epidermis (VE), as well as detailed information concerning the keratinocyte network, including nuclei size, shape, and density. 11 , 23 Additionally, these algorithms enable the evaluation of cellular network atypia 24 and superficial dermal thickness. 23

An initial study conducted by Chauvel‐Picard et al. demonstrated the technique's ability to observe variations in the thickness of SC and VE and to quantify modifications in cellular metrics within the living epidermis across different body sites. 11 A separate study by Breugnot et al. focused on examining the superficial dermis to correlate structural features of collagen with aging, but no further study of superficial layers was intended. 25 In 2023, Bonnier et al. reported an exploratory study that included 100 Caucasian female healthy volunteers to investigated facial skin aging coupling LC‐OCT and AI based images analysis protocols. 26 However, as of now, the literature available for characterising healthy skin and studying the effects of aging remains limited, despite the high expectations of advancing our knowledge through high‐performance imaging technologies. This is especially important to address the various structural and functional variations between skin types, such as those between Caucasians and Asians, which have garnered significant interest within the fields of dermatology and the cosmetic industry.

In terms of macroscopic, i.e. perceived, signs of aging, it has been observed that Caucasian individuals tend to develop fine lines and wrinkles earlier and more prominently than Asians, while Asian skin is typically thicker than Caucasian skin, making it less prone to sagging and fine lines. In contrast, Asian skin is more prone to pigmentation disorders, such as melasma or hyperpigmentation, which can be triggered by UV exposure, whereas Caucasians may experience freckles and age spots as they age. Other features characterizing Asian skin, such as a lower production of sebum (skin oil) contributing to a more matte appearance, may be associated with a lower incidence of acne in Asian populations. Additionally, Asian skin has smaller pores compared to Caucasian skin, affecting the appearance of skin texture and susceptibility to acne. Higher levels of natural hydration in Asian skin can contribute to a plump and youthful appearance. 27 , 28

In terms of age‐related modifications in subsurface microstructures of the skin, there is a limited number of studies in the literature addressing the comparison of aging between Caucasian and Asian skin at the histological level. 29 Notably, the face is commonly regarded as the central aspect of one's appearance, carrying substantial social and cultural significance, thereby influencing the self‐perception of aging faces and impacting self‐esteem and confidence. It is crucial to conduct more comprehensive investigations into the modifications in skin layers due to aging, especially in the context of comparisons between Asian and Caucasian populations. This can be facilitated by leveraging the capabilities of recently developed technologies, which allow more convenient data collection from sensitive skin areas, such as the facial regions in women.

The present study represents the first extensive investigation utilising LC‐OCT 3D imaging to examine the effects of facial skin aging in healthy Asian women. This was achieved through the aid of AI‐assisted computations to derive quantitative histological and cellular biomarkers from a cohort of 109 volunteers, ranging in age from 20 to 70 years. The acquisition of LC‐OCT 3D images encompassed three facial regions: the temple, cheekbone, and mandible. Subsequently, the analysis included assessments of skin layer thickness (stratum corneum and viable epidermis), evaluations of the undulation of the DEJ, and detailed examinations of cell morphology, encompassing keratinocyte nuclei size and shape, number of cell layers, and the presence of atypia. Furthermore, the results obtained from Asian females in this study were compared with those obtained from 100 healthy Caucasian female volunteers, as previously documented in ref. 26, with the aim of discerning variations in age‐related facial skin modifications between these two ethnic groups. Consequently, in addition to presenting an expanded dataset of LC‐OCT data from a substantial cohort of healthy females, this study represents the initial quantitative comparison of histological and cellular metrics between ethnic groups, specifically composed of Caucasian and Asian females.

2 MATERIALS AND METHODS

2.1 Study population

The study was conducted on 109 healthy Asian female volunteers in China, under the oversight of the Shanghai Ethics Committee for Clinical Research (SECCR/2022‐77‐01), following strict adherence to the principles of the Declaration of Helsinki. All participants provided written informed consent, reinforcing the ethical and regulatory framework governing these investigations. Participants were recruited to form five evenly distributed age groups, spanning the ranges of [20‐30], [31‐40], [41‐50], [51‐60], and [61‐70] years old. All volunteers participated for 1 day, during which LC‐OCT imaging was performed. The results obtained from LC‐OCT images of healthy Asian females were compared to a previously reported study on female Caucasians conducted in France, which involved one hundred healthy Caucasian female volunteers and was supervised by the French ethical committee ‘Comité de protection des personnes sud méditerranée I’ (IDRCB: 2021‐A00101‐40). Further details for the study conducted on Caucasian females can be found in reference. 26

2.2 LC‐OCT imaging

A DeepLive system (DAMAE MEDICAL, France) was employed to capture images, facilitating the in vivo 3D visualization of the skin, as illustrated in Figure 1. This system utilizes a two‐beam Linnik interference microscope equipped with a supercontinuum laser that emits light at a central wavelength of 800 nm. This configuration allows for the acquisition of tomographic images both perpendicular and parallel to the skin's surface, achieving a rate of eight frames per second. The isotropic lateral and axial resolutions provide cellular scale. 17 To ensure optimal imaging conditions, paraffin oil was used as an immersion medium, chosen for its matching refractive index with that of the skin (n ∼ 1.4). For the purposes of this study, a total of three LC‐OCT 3D images was acquired per subject, each measuring 1200 µm × 500 µm × 350 µm. These images were obtained as stacks of parallel slices relative to the skin surface, with a step size of 1 µm in the z‐direction. Importantly, any pigmented spots were deliberately excluded from the image acquisitions. The LC‐OCT acquisitions were conducted on three distinct facial regions: the cheekbone, temple, and the lower jawline (mandible). To ensure unbiased assessments, a randomisation method was applied to determine whether the left or right side of the face was assessed for each subject. Between the acquisition of each 3D image, the probe was displaced by approximately 1–2 mm to facilitate comprehensive coverage of the target areas. The dataset subjected to analysis in these studies comprised a total 981 3D LC‐OCT images. It should be noted that the results from the study conducted on 100 female Caucasian volunteers used for the current comparison of the effects of aging represented an equivalent dataset in terms of size, with 900 3D LC‐OCT images. 26

FIGURE 1 Example of LC‐OCT images on healthy skin. (A) 3D stack, (B) 2D horizontal image (xy) and (C) 2D reconstructed vertical image (xz).

2.3 LC‐OCT images analysis

2.3.1 Thickness of skin layers and DEJ undulation

The segmentation process involved several steps. Initially, 2D U‐net models were utilised to segment individual reconstructed vertical images (as shown in Figure 1C). This segmentation allowed for the classification of pixels into three distinct categories, representing different skin interfaces: pixels situated below the skin surface, below the SC, and below the DEJ, as illustrated in Figure 2. To achieve a comprehensive 3D result, the segmentation algorithm was applied to all vertical views in both the xz and yz directions. Subsequently, to ensure accuracy, the results of the segmentations for the 3D LC‐OCT images were meticulously reviewed by a trained operator. Any regions corresponding to hair follicles were systematically excluded from the calculation of the metrics. The metrics for SC thickness and VE thickness were derived from the average number of pixels between the segmented skin surface and the segmented SC interface, as well as between the segmented SC and the segmented DEJ, as visualized in Figure 2.

FIGURE 2 Segmentation of layers. (A) Skin surface, (B) SC‐Viable epidermis interface and (C) Epidermis‐Dermis interface (i.e., dermal‐ epidermal junction).

The DEJ undulation was defined as the surface area of the dermo‐epidermis interface normalised by the area of the analysed field of view. 11 The DEJ undulation index was calculated as followed 30 : Udej=SDEJ/SROI−1,

with SDEJ being the area of the DEJ layer and SROI being the total horizontal area of the LC‐OCT image without areas corresponding to hair follicles.

2.3.2 Cellular metrics

To segment keratinocytes, a deep learning model, specifically an AI algorithm based on 3D convolutions, was employed, as previously described in detail. 11 The chosen architecture for this task was the 3D StarDist model, 31 primarily because it excels at detecting 3D star‐convex shapes at the instance level, which closely resembles the structure of keratinocyte nuclei. It's important to note that this deep learning model, based on a variant of a 3D ResNet18, was trained from scratch. This was necessitated by the absence of any pretrained weights available for a similar task at the time of implementation. The 3D StarDist model leverages 3D convolutions to predict, at the voxel level (which corresponds to a 3D pixel), both the probability of a voxel being the centre of a nucleus and the lengths of various radii that define the nucleus. In total, 96 radii were employed, determined from a 3D Fibonacci lattice, ensuring a comprehensive account of the diversity in 3D nuclei shapes and sizes. To achieve accurate cell detection, a probability threshold of 0.5 was set for cell detection, and a Non‐Maximum Suppression threshold of 0.05 was applied. This configuration enabled the detection of touching but non‐overlapping cells within the images. For further details and a more comprehensive explanation of the methodology, additional information can be found in the study authored by Chauvel‐Picard et al. 11

The Cell Surface Density (CSD), expressed as cell number per square millimetre (cell number/mm2), is determined by dividing the total count of nuclei in the epidermis by the surface area of the skin. The Number of Cell Layers (NCL) was calculated by averaging the number of cells traversed along the vertical axis between the stratum corneum/stratum granulosum (SC/SG) junction and the DEJ interface. Nucleus Volume (NV), expressed in cubic micrometres (µm3), was computed using star‐convex polyhedra to detect and segment cell nuclei. Nucleus Compactness (NC) serves as an indicator of sphericity and is determined based on the 3D surface area of the nucleus (A) and the cell volume (V). The formula used is (36ΠV2/A3), and compactness is expressed as a score ranging from 0 to 1, where 1 signifies a perfect sphere. Cell Network Atypia (CNA) involves identifying outliers within the cell population using features derived from nuclei segmentation. This results in binary scores classifying nuclei as either atypical or not. The algorithm used for this purpose was trained using samples from both pathological and healthy skin conditions, with the assumption that atypical nuclei are more prevalent during pathological processes such as actinic keratosis or Squamous cell carcinoma. Further details about the model employed, XGBoost, 32 can be found in ref. 24.

As an additional marker of aging, the standard deviation of the metrics NCL, NV, and NC was calculated and utilised as quantitative measures. The standard deviation was computed for each 3D stack individually and employed as a representative mean of the metrics heterogeneity to assess their correlation with aging. These standard deviations are denoted as STD_NCL, STD_NV, and STD_NC.

2.4 Statistical analysis

Statistical analysis was conducted using MATLAB (MathWorks, USA). Quantitative variables were summarized using means and standard deviations. To assess differences among subjects within different age groups for each quantitative variable characterizing skin features, an analysis of variance (ANOVA) was executed. The normality of the residuals from the ANOVA was checked using a Shapiro–Wilk test, with a significance level set at 10%. In cases where the residuals did not exhibit normal distribution, a Kruskal–Wallis test was utilized as an alternative. When the ANOVA or Kruskal–Wallis test yielded a significant result, a post‐hoc test was subsequently employed. This post‐hoc test allowed for pairwise comparisons between age groups, with Tukey's adjustment applied to accommodate for multiple comparisons and maintain statistical rigor. In addition, when comparing ethnicity groups within each age group, the normality of the data within each subgroup was evaluated using the Shapiro–Wilk test prior to analysis. A p‐value greater than 0.05 was considered indicative of a normal distribution. Subsequently, if both groups exhibited a normal distribution, the Student's t‐test was employed to compare means. Conversely, in cases where the data deviated from normality (Shapiro–Wilk test p ≤ 0.05), the non‐parametric Wilcoxon rank‐sum test was applied.

3 RESULTS

3.1 Histological metrics

3.1.1 Effects of ageing

Table 1 presents the mean SC thickness, mean VE thickness, and the undulation of the DEJ for Asian female volunteers in the temple, cheekbone, and mandible regions. Significant differences were observed in SC thickness for the temple (p = 0.028) and the mandible (p < 0.001). Specifically, in the temple region, the most notable variation was between the [20‐30] age group (13.2 ± 0.6 µm) and the [51‐60] age group (13.7 ± 0.8 µm), representing a 3.8% increase. For the mandible, the most significant difference was observed between the [20‐30] age group (13.6 ± 0.5 µm) and the [51‐60] age group (14.3 ± 0.7 µm), constituting a 5.1% increase (p < 0.001). In comparison to histological metrics obtained from the previous study conducted on Caucasian females, 26 the temple exhibited a 7.6% increase in SC thickness between the [20‐30] age group (13.2 ± 0.8 µm) and the [61‐70] age group (14.2 ± 1.0 µm). For the mandible, a 6.1% increase was observed between the [20‐30] age group (13.1 ± 0.4 µm) and the [61‐70] age group (13.9 ± 1.3 µm). Although for both Caucasian and Asian cohorts the thickness of the SC tends to increase, the effect appeared more correlated with aging in the Caucasian females while for Asian females the maximum mean thickness is reached at an earlier stage, for instance for the [51‐60] group. Notably, no significant age‐related variations in SC thickness were observed in either age group for the cheekbone region (Table 1, 26 ). When it comes to the thickness of the VE, no significant age‐related variations were observed in either facial area (Table 1). In contrast, for Caucasian females a noticeable thinning with age was noted in the mandible region, amounting to a 12.7% decrease from 55.1 ± 7.2 µm (in the [20‐30] age group) to 48.1 ± 5.3 µm (in the [61‐70] age group). This decrease was statistically significant (Kruskal–Wallis p = 0.0003). For the temple and cheekbone regions, similar to the Asian cohort, no significant variations were observed in this regard. The undulation of the DEJ did not exhibit any significant changes with age for either panel or across any of the three facial areas examined (p > 0.05).

TABLE 1 SC thickness, VE thickness and DEJ undulation determined for the temple, cheekbone and mandible from LC‐OCT 3D images collected from Asian healthy female volunteers.

	SC thickness (µm)	VE thickness (µm)	Undulation DEJ (%)	
Temple							
[20,30]	13.2 ± 0.6	a	55.1 ± 8.0	–	7.33 ± 8.25	–	
[31,40]	13.2 ± 0.4	ab	53.9 ± 9.0	–	8.03 ± 8.92	–	
[41,50]	13.4  ± 0.4	ab	57.3  ± 7.4	–	12.0  ± 11.4	–	
[51,60]	13.7  ± 0.8	b	57.3  ± 7.5	–	7.36  ± 6.50	–	
[61,70]	13.6  ± 0.7	ab	55.4  ± 6.5	–	9.21  ± 4.61	–	
	p = 0.028 (KW)	p = 0.54 (A)	p = 0.074 (KW)	
Cheekbone							
[20,30]	13.4  ± 0.5	–	55.8  ± 9.2	–	4.46  ± 5.10	–	
[31,40]	13.2  ± 0.3	–	52.7  ± 7.3	–	3.20  ± 4.11	–	
[41,50]	13.9  ± 2.0	–	56.5  ± 12	–	6.40  ± 8.58	–	
[51,60]	13.6  ± 0.7	–	52.0  ± 7.1	–	3.67  ± 3.06	–	
[61,70]	13.6  ± 0.8	–	53.3  ± 9.7	–	3.98  ± 2.95	–	
 	p = 0.244 (KW)	p = 0.457 (KW)	p = 0.822 (KW)	
Mandible	 	 	 	 	 	 	
[20,30]	13.6  ± 0.5	ac	54.5  ± 6.9	–	7.56  ± 5.72	–	
[31,40]	13.5  ± 0.5	a	57.3  ± 7.9	–	9.14  ± 7.15	–	
[41,50]	13.9  ± 0.9	abc	56.4  ± 12	–	12.7  ± 11.8	–	
[51,60]	14.3  ± 0.7	b	53.8  ± 6.4	–	9.42  ± 6.86	–	
[61,70]	14.1  ± 0.7	bc	56.6  ± 8.5	–	10.4  ± 7.18	–	
 	p < 0.001 (KW)	p = 0.497 (KW)	p = 0.511 (KW)	
Note: Letters indicate the outcome from multi‐comparison tests (when relevant).

Abbreviations: A, ANOVA; KW, Kruskal–Wallis.

John Wiley & Sons, Ltd.

3.1.2 Comparison of ethnicity groups

Table 2 presents a statistical analysis between facial histological metrics obtained from each age group of the Asian and Caucasian cohorts. The thickness of the SC at the temple exhibited a significantly lower mean (p = 0.025) in the Asian [61‐70] group (13.6 ± 0.7 µm) compared to the same age group of Caucasian female volunteers (14.2 ± 1.0 µm). In contrast, for the cheekbone, the Asian cohort showed a significantly thicker SC for all age groups (p < 0.001), with a mean increase of 9.6%. The most significant difference was observed in the age group [41‐50], with mean values of 13.9 ± 2.0 µm for the Asian group and 12.5 ± 0.9 µm for the Caucasian group. For the mandible, a thicker SC was also observed in Asian females for the age groups [20‐30], [41‐50], and [51‐60], with increases of 3.4%, 4.2%, and 4.4%, respectively.

TABLE 2 Statistical analysis applied to SC thickness, VE thickness and DEJ undulation determined for the temple, cheekbone and mandible from LC‐OCT 3D images collected from Caucasian and Asian female volunteers.

		SC thickness (µm)	VE thickness (µm)	Undulation DEJ (%)	
Temple	[20,30]	p = 0.93

(Wilcoxon)

	–	p = 0.82

(t‐student)

	–	p = 0.029

(Wilcoxon)

	177.5% ↗	
	[31,40]	p = 0.68

(Wilcoxon)

	–	p = 1

(Wilcoxon)

	–	p = 0.013

(Wilcoxon)

	247.0% ↗	
	[41,50]	p = 0.41

(Wilcoxon)

	–	p < 0.001

(Wilcoxon)

	11.4% ↗	p < 0.001

(Wilcoxon)

	464.7% ↗	
	[51,60]	p = 0.67

(Wilcoxon)

	–	p = 0.034

(t‐student)

	8.1% ↗	p = 0.001

(Wilcoxon)

	204.3% ↗	
	[61,70]	p = 0.025

(Wilcoxon)

	4.5% ↘	p = 0.07

(t‐student)

	–	p < 0.001

(Wilcoxon)

	344.4% ↗	
Cheekbone	[20,30]	p < 0.001

(Wilcoxon)

	10.7% ↗	p = 0.004

(t‐student)

	15.4% ↗	p < 0.001

(Wilcoxon)

	1629.7% ↗	
	[31,40]	p < 0.001

(t‐student)

	6.6% ↗	p = 0.09

(t‐student)

	–	p < 0.001

(Wilcoxon)

	593.4% ↗	
	[41,50]	p < 0.001

(Wilcoxon)

	11.7% ↗	p < 0.001

(Wilcoxon)

	26.4% ↗	p < 0.001

(Wilcoxon)

	1642.2% ↗	
	[51,60]	p < 0.001

(Wilcoxon)

	10.4% ↗	p = 0.015

(t‐student)

	11.1% ↗	p < 0.001

(Wilcoxon)

	671.4% ↗	
	[61,70]	p < 0.001

(Wilcoxon)

	8.6% ↗	p = 0.003

(Wilcoxon)

	16.7% ↗	p < 0.001

(Wilcoxon)

	726.5% ↗	
Mandible	[20,30]	p = 0.002

(t‐student)

	3.4% ↗	p = 0.93

(Wilcoxon)

	–	p < 0.001

(Wilcoxon)

	425.7% ↗	
	[31,40]	P = 0.12

(Wilcoxon)

	–	p = 0.93

(t‐student)

	–	p < 0.001

(Wilcoxon)

	670.7% ↗	
	[41,50]	P = 0.0037

(t‐student)

	4.2% ↗	p = 0.16

(Wilcoxon)

	–	p < 0.001

(Wilcoxon)

	795.1% ↗	
	[51,60]	p = 0.001

(Wilcoxon)

	4.4% ↗	p = 0.33

(Wilcoxon)

	–	p < 0.001

(Wilcoxon)

	556.0% ↗	
	[61,70]	p = 0.34

(Wilcoxon)

	–	p < 0.001

(t‐student)

	17.5% ↗	p < 0.001

(Wilcoxon)

	857.2% ↗	
Note: When significant, the arrows represent the variation observed in Asian female as compared to Caucasian females.

John Wiley & Sons, Ltd.

The thickness of the VE for the temple was found to be significantly higher in the Asian group for the age groups [41‐50] (p < 0.001) and [51‐60] (p = 0.034). There was a substantial increase of approximately 11.4% and 8.1% for the [41‐50] and [51‐60] age groups, respectively, when comparing Asian and Caucasian females in the temple region. For the cheekbone, there was a significant increase in thickness among Asian females for age groups [20‐30], [41‐50], [51‐60], and [61‐70], ranging from approximately 10% to about 21%. Within each age range studied, a noteworthy increase of an average of 15.4% was observed between Asian and Caucasian females in the cheekbone region. Regarding the mandible, no difference was observed between Asian and Caucasian females across all age groups. However, there was a noticeable increase of about 17.5% for the [61‐70] age group among Asian females, with an average VE thickness of 56.6 ± 8.5 µm, while Caucasian females showed an average VE thickness of 48.1 µm.

The percentage of undulation in DEJ exhibited significant differences between Asian and Caucasian females, irrespective of facial region or age group. Mean values were found to be 287.6% (p < 0.05), 1052.6% (p < 0.001), and 660.9% (p < 0.001) higher in Asian females compared to Caucasian females for the temple, cheekbone, and mandible, respectively.

3.2 Cellular metrics

3.2.1 Effects of ageing

Number of cell layers:

The mean NCL and their respective standard deviations, computed for the three facial areas under investigation, are presented in Table 3. In the case of Asian females, no significant variations were observed in these metrics. This aligns with the consistent thickness of the VE observed across different age groups, as shown in Table 1. In contrast, for Caucasian females, 26 the mean NCL in the mandible decreased from 6.28 ± 0.65 layers (in the [20‐30] age group) to 5.64 ± 0.57 layers (in the [61‐70] age group) (p = 0.002). Additionally, the STD_NCL also exhibited a significant decrease, from 1.19 ± 0.12 to 1.05 ± 0.11 layers (p < 0.001). Pairwise comparisons revealed distinct differences between the youngest and oldest age groups, indicating a higher level of variability in the NCL in the [20‐30] age group. The observed decrease of approximately 7 µm in the VE thickness for the mandible correlated with the reduction in the NCL. The decreased standard deviation indicated a more uniform and consistent appearance. For the temple and cheekbone regions, the NCL did not exhibit any significant differences. However, it is worth noting that, for the temple, the STD_NCL confirmed a higher level of variability in the youngest group (p = 0.038). 26

TABLE 3 Number of cell layers (NCL) and standard deviation for the number of cell layers (STD_NCL) determined for the temple, the cheekbone and the mandible.

	NCL	STD_NCL	
Temple					
[20,30]	6.33 ± 0.57	–	1.25 ± 0.21	–	
[31,40]	6.26 ± 0.75	–	1.24 ± 0.26	–	
[41,50]	6.51 ± 0.71	–	1.34 ± 0.21	–	
[51,60]	6.48 ± 0.64	–	1.27 ± 0.14	–	
[61,70]	6.39 ± 0.56	–	1.23 ± 0.16	–	
 	p = 0.718 (A)	p = 0.435 (A)	
Cheekbone	 	 	 	 	
[20,30]	6.20 ± 0.76	–	1.26 ± 0.23	–	
[31,40]	6.06 ± 0.63	–	1.16 ± 0.17	–	
[41,50]	6.36 ± 0.99	–	1.25 ± 0.35	–	
[51,60]	6.03 ± 0.64	–	1.15 ± 0.14	–	
[61,70]	6.09 ± 0.82	–	1.18 ± 0.24	–	
	p = 0.635 (A)	p = 0.507 (KW)	
Mandible					
[20,30]	6.38 ± 0.63	–	1.24 ± 0.19	–	
[31,40]	6.57 ± 0.68	–	1.27 ± 0.20	–	
[41,50]	6.45 ± 0.80	–	1.31 ± 0.35	–	
[51,60]	6.40 ± 0.50	–	1.22 ± 0.15	–	
[61,70]	6.49 ± 0.71	–	1.27 ± 0.25	–	
 	p = 0.809 (KW)	p = 0.964 (KW)	
Abbreviations: A, ANOVA; KW, Kruskal–Wallis.

John Wiley & Sons, Ltd.

Nuclei morphology:

Table 4 presents the results of the statistical analysis performed on the CSD, NV, STD_NV, NC, STD_NC, and CNA for the temple, cheekbone, and mandible regions in both Asian and Caucasian female volunteers. The provided p values for each metric were calculated based on data from five age groups. p values less than 0.05 are highlighted in blue. The standard deviation (STD) calculated for volume and compactness metrics reflects the variation within individual LC‐OCT 3D images. For each facial area and each metric, the results obtained from Caucasian and Asian females are presented side by side for comparison. Detailed metrics (mean ± standard deviation) can be found in supplementary materials in Table S1 (CSD), Table S2 (NV and STD_NV), Table S3 (NC and STD_NC), and Table S4 (CNA).

TABLE 4 Summary of statistical analysis performed on cellular metrics for the temple, the cheekbone and the mandible.

	Temple	Cheekbone	Mandible	
	Caucasian	Asian	Caucasian	Asian	Caucasian	Asian	
CSD (cells /mm2)	p = 0.019 (KW)	p = 0.929 (A)	p = 0.701 (A)	p = 0.9 (A)	p < 0.001 (A)	p = 0.53 (KW)	
↘	–	–	–	↘	–	
NV (µm3)	p = 0.007 (KW)	p = 0.021 (KW)	p = 0.334 (A)	p = 0.044 (KW)	p = 0.013 (KW)	p < 0.001 (A)	
	↗	↗	–	↗	↗	↗	
STD_NV (µm3)	p = 0.0006 (KW)	p = 0.007 (KW)	p = 0.002 (A)	p = 0.002 (KW)	p < 0.001 (KW)	p < 0.001 (KW)	
	↗	↗	↗	↗	↗	↗	
NC	p = 0.036 (KW)	p = 0.22 (KW)	p = 0.001 (A)	p = 0.008 (A)	p = 0.044 (KW)	p = 0.92 (A)	
	–	–	↘	↘	↘	–	
STD_NC	p = 0.108 (KW)	p = 0.881 (A)	p = 0.232 (A)	p = 0.141 (A)	p = 0.628 (KW)	p = 0.454 (KW)	
	–	–	–	–	–	–	
CNA	p < 0.001 (KW)	p = 0.005 (KW)	p = 0.002 (A)	p = 0.005 (KW)	p < 0.001 (KW)	p = 0.06 (A)	
	↗	↗	↗	↗	↗	(↗)	
Note: p values for each metric are provided. Highlighted in blue are metrics with p < 0.05. Symbols “↗”, “↘”, “–” indicate the evolution of mean values according to aging.

Abbreviations: A, ANOVA; CNA, cell network atypia; CSD, cell surface density; KW, Kruskal–Wallis; NC, nuclei compactness; NV, nuclei volume; STD, standard deviation computed as a metric.

John Wiley & Sons, Ltd.

CSD displayed a significant decrease for the temple and mandible of Caucasian females (p < 0.05, Table 4). The variations observed were from 30306 ± 3944 cells/mm2 to 26021 ± 3663 cells/mm2 and from 31641 ± 4247 cells/mm2 to 26835 ± 3511 cells/mm2, age group [20‐30] and [61‐70], respectively. In comparison no age‐related variations were observed for Asian females.

NV exhibited significant age‐related increase for both Caucasian (temple and mandible) and Asian females (temple, cheekbone and mandible) (Table 4). For Caucasian females the variations ranged from 143.5 ± 5.8 mm3 to 150.5 ± 9.3 mm3 and 148.3 ± 9.7 mm3 to 158.1 ± 11.1 mm3 between age group [20‐30] and [61‐70], corresponding to an increase of 5% and 7% for the temple and the mandible, respectively, Similarly, for Asian females, the temple displayed an increase from 146.5 ± 5.5 mm3 to 151.8 ± 6.1 mm3 (+4%), the cheekbone an increase from 148.7 ± 6.9 mm3 to 155.2 ± 10.1 mm3 (+4%) and the mandible from 144.8 ± 5.5 mm3 to 154.7 ± 6.2 mm3 (+6%). The STD_NV displayed the same consistency between the two ethnics studied. For Caucasian females the mean STD_NV increased from 67.5 ± 3.4 mm3 to 73.3 ± 6.8 mm3 for the temple (+5%), 71.6 ± 4.7 mm3 to 78.2 ± 8.1 mm3 for the cheekbone (+8%) and 68.8 ± 5.1 mm3 to 78.9 ± 9.5 mm3 for the mandible (+14%), compared to, for Asian females, an increase from 66.6 ± 3.6 mm3 to 72.1 ± 5.5 mm3 for the temple (+9%), 67.1 ± 5.3 mm3 to 72.5 ± 6.2 mm3 for the cheekbone (+9%) and 67.1 ± 3.8 mm3 to 74.3 ± 5.7 mm3 for the mandible (+10%).

A significant decrease for NC was found for the cheekbone (−2%) and the mandible (−1%) of Caucasian females, corresponding to variations from 0.771 ± 0.012 to 0.754 ± 0.014 and 0.776 ± 0.012 to 0.771 ± 0.011, respectively. Despite the p = 0.036 found for the temple, the multi‐comparison test showed no significant variations between age group [20‐30] and [61‐70]. For Asian females, the decrease from 0.783 ± 0.010 to 0.771 ± 0.012 for the cheekbone (−2%) was the only variation observed. The STD_NC didn't display any variations that could be correlated with ageing.

The CNA displayed an increase for Caucasian and Asian females for the three facial areas studied. For Caucasian females, the variations observed were from 0.163 ± 0.023 to 0.203 ± 0.036 for the temple (+25%), from 0.207 ± 0.025 to 0.241 ± 0.044 for the cheekbone (+17%) and from 0.169 ± 0.033 to 0.223 ± 0.052 for the mandible (+32%). For female Asians significant increases (p < 0.05) for the temple, with means equal to 0.148 ± 0.021 and 0.174 ± 0.022 (+15%), and the cheekbone with 0.153 ± 0.029 and 0.179 ± 0.031 (+15%), were observed for the age groups [20‐30] and [61‐70], respectively. For the mandible, despite a p value of 0.06, the means values displayed a similar trend with age group [20‐30] = 0.166 ± 0.026 and age group [61‐70] = 0.189 ± 0.038 (+14%) (Table 4).

3.2.2 Comparison of ethnicity groups

Table 5 provides a comparison of quantitative cellular metrics between Asian and Caucasian females across different age groups and facial regions.

TABLE 5 Statistical analysis comparing the number of cell layers (NCL), the standard deviation of the number of cell layers (STD_NCL), and the cell surface density (CSD) for Asian females against Caucasian females.

	NCL	STD_NCL	CSD (cells /mm2)	
Temple							
[20,30]	p = 0.33

(t‐student)

	–	p = 0.55

(Wilcoxon)

	–	p = 0.21

(t‐student)

	–	
[31,40]	p = 0.24

(t‐student)

	–	p = 0.94

(t‐student)

	–	p = 0.011

(Wilcoxon)

	13.3% ↗	
[41,50]	p = 0.001

(Wilcoxon)

	11.0% ↗	p = 0.002

(t‐student)

	13.8% ↗	p = 0.001

(t‐student)

	15.9% ↗	
[51,60]	p = 0.002

(t‐student)

	10.1% ↗	p = 0.06

(Wilcoxon)

	–	p < 0.001

(t‐student)

	20.4% ↗	
[61,70]	p = 0.003

(t‐student)

	8.9% ↗	p = 0.04

(t‐student)

	7.8% ↗	p < 0.001

(Wilcoxon)

	24.6% ↗	
Cheekbone							
[20,30]	p < 0.001

(t‐student)

	15.4% ↗	p = 0.01

(t‐student)

	14% ↗	p < 0.001

(t‐student)

	20.7% ↗	
[31,40]	p = 0.02

(t‐student)

	8.5% ↗	p = 0.26

(t‐student)

	–	p = 0.005

(t‐student)

	14.6% ↗	
[41,50]	p < 0.001

(t‐student)

	21.1% ↗	p = 0.006

(Wilcoxon)

	22.9% ↗	p < 0.001

(t‐student)

	22.4% ↗	
[51,60]	p < 0.001

(t‐student)

	13% ↗	p = 0.02

(t‐student)

	9.1% ↗	p < 0.001

(t‐student)

	17.5% ↗	
[61,70]	p = 0.004

(Wilcoxon)

	13.4% ↗	p = 0.04

(Wilcoxon)

	14.8% ↗	p = 0.001

(t‐student)

	20.6% ↗	
Mandible							
[20,30]	p = 0.6

(t‐student)

	–	p = 0.36

(t‐student)

	–	p = 0.86

(t‐student)

	–	
[31,40]	p = 0.36

(t‐student)

	–	p = 0.4

(t‐student)

	–	p = 0.66

(t‐student)

	–	
[41,50]	p = 0.09

(Wilcoxon)

	–	p = 0.04

(Wilcoxon)

	17.5% ↗	p = 0.79

(Wilcoxon)

	–	
[51,60]	p = 0.005

(Wilcoxon)

	11.1% ↗	p = 0.008

(Wilcoxon)

	11.1% ↗	p = 0.004

(t‐student)

	12.9% ↗	
[61,70]	p < 0.001

(t‐student)

	15.0% ↗	p < 0.001

(Wilcoxon)

	21.5% ↗	p = 0.001

(t‐student)

	15.6% ↗	
Note: p‐values for each metric are provided. Metrics with p < 0.05 are highlighted in blue. Symbols “↗”, “↘”, “–” indicate the evolution of mean values in Asians volunteers as compared to Caucasian females.

John Wiley & Sons, Ltd.

On the temple, NCL was significantly higher (p < 0.005), approximately 8%–11%, in Asian females for the age groups [41‐50], [51‐60], and [61‐70], respectively. Simultaneously, the STD_NCL was also significantly higher (p < 0.05) for the age groups [41‐50] and [61‐70]. The age group [51‐60] displayed a similar trend, but the p‐value of 0.06 didn't reach statistical significance. On the cheekbone, the NCL was consistently found to be significantly higher in Asian females (p < 0.05), with an increase ranging from 8.5% (age group [31‐40]) to 21.1% (age group [41‐50]). The STD_NCL exhibited a similar pattern, except for the age group [31‐40]. On the mandible, the age groups [51‐60] and [61‐70] exhibited a 11.11% (p = 0.005) and 15% (p < 0.001) increase, respectively. As observed in other facial areas, the STD_NCL followed a similar trend, with an increase ranging from approximately 11.1% for the age group [51‐60] to around 21.5% for the age group [61‐70]. It's worth noting that the age group [41‐50] also displayed a significant increase of 17.5%. These variations between Asian and Caucasian females indicate a higher level of heterogeneity in the NCL within the Asian population. The NCL and the thickness of the viable epidermis appeared to be correlated to some extent when comparing Asian and Caucasian females, notably for the cheekbone regardless of age groups.

The CSD was found to be significantly higher on the temple and the cheekbone, with increases ranging from approximately 13% (p = 0.011) to around 24% (p < 0.001) for the temple and about 14% (p = 0.005) to roughly 22% (p < 0.001) for the cheekbone. Only the age group [20‐30] for the temple didn't exhibit significant variation. For the mandible, the differences were not as consistent across age groups, with a significant increase of approximately 13% (p = 0.004) and roughly 16% (p = 0.001) observed for age groups [51‐60] and [61‐70], respectively. This reduction in CSD between Asian and Caucasian females was also correlated with the decrease in VE thickness.

With regard to NV and STD_NV (Table 6), no significant variations were observed for the temple. For the cheekbone, age groups [20‐30] and [31‐40] displayed a significant decrease in NV, approximately 4%–5%. However, only the age group [20‐30] exhibited a significant decrease of around 7% in the STD_NV (p < 0.001) in Asian females. The age group [61‐70] also showed a significant decrease of about 8% (p = 0.01). For the mandible, age groups [31‐40] and [41–50] had a significant decrease in NV, while only the age group [51‐60] had a significant decrease in STD_NV.

TABLE 6 Statistical analysis comparing the cellular metrics for Asian females against Caucasian females.

	NV (µm3)	STD_NV (µm3)	NC	STD_NC	CNA	
Temple											
[20,30]	p = 0.1

(t‐student)

	–	p = 0.37

(t‐student)

	–	p = 0.51

(t‐student)

	–	p = 0.82

(t‐student)

	–	p = 0.02

(t‐student)

	9.1% ↘	
[31,40]	p = 0.72

(t‐student)

	–	p = 0.93

(t‐student)

	–	p = 0.64

(t‐student)

	–	p = 0.59

(t‐student)

	–	p = 0.009

(Wilcoxon)

	8.8% ↘	
[41,50]	p = 0.88

(Wilcoxon)

	–	p = 0.1

(Wilcoxon)

	–	p = 0.008

(Wilcoxon)

	1.1% ↗	p = 0.22

(Wilcoxon)

	–	p < 0.001

(Wilcoxon)

	18.4% ↘	
[51,60]	p = 0.12

(Wilcoxon)

	–	p = 0.77

(Wilcoxon)

	–	p = 0.35

(t‐student)

	–	p = 0.47

(Wilcoxon)

	–	p = 0.01

(Wilcoxon)

	10.9% ↘	
[61,70]	p = 0.14

(Wilcoxon)

	–	p = 0.53

(Wilcoxon)

	–	p = 0.02

(t‐student)

	0.8% ↘	p = 0.38

(t‐student)

	–	p = 0.006

(Wilcoxon)

	14.4% ↘	
Cheekbone											
[20,30]	p = 0.001

(t‐student)

	4.8% ↘	p < 0.001

(Wilcoxon)

	6.3% ↘	p = 0.002

(t‐student)

	1.4% ↗	p < 0.001

(t‐student)

	10.0% ↘	p < 0.001

(Wilcoxon)

	26.1% ↘	
[31,40]	p = 0.01

(t‐student)

	3.9% ↘	p = 0.07

(t‐student)

	–	p < 0.001

(Wilcoxon)

	1.3% ↗	p < 0.001

(t‐student)

	8.8% ↘	p < 0.001

(t‐student)

	17.3% ↘	
[41,50]	p = 0.82

(t‐student)

	–	p = 0.85

(Wilcoxon)

	–	p < 0.001

(t‐student)

	2.7% ↗	p = 0.002

(t‐student)

	9.0% ↘	p < 0.001

(Wilcoxon)

	19.3% ↘	
[51,60]	p = 0.08

(Wilcoxon)

	–	p = 0.18

(Wilcoxon)

	–	p < 0.001

(t‐student)

	1.9% ↗	p < 0.001

(t‐student)

	8.4% ↘	p < 0.001

(Wilcoxon)

	18.6% ↘	
[61,70]	p = 0.12

(t‐student)

	–	p = 0.01

(t‐student)

	7.2% ↘	p < 0.001

(t‐student)

	2.1% ↗	p = 0.002

(t‐student)

	8.6% ↘	p < 0.001

(t‐student)

	25.9% ↘	
Mandible											
[20,30]	p = 0.41

(Wilcoxon)

	–	p = 0.23

(Wilcoxon)

	–	p = 0.88

(Wilcoxon)

	–	p = 0.01

(t‐student)

	5.2% ↘	p = 0.66

(Wilcoxon)

	–	
[31,40]	p = 0.01

(Wilcoxon)

	4.1% ↘	p = 0.07

(t‐student)

	–	p = 0.52

(Wilcoxon)

	–	p = 0.002

(t‐student)

	5.7% ↘	p = 0.17

(Wilcoxon)

	–	
[41,50]	p = 0.04

(t‐student)

	2.7% ↘	p = 0.02

(t‐student)

	4.3% ↘	p = 0.003

(t‐student)

	1.4% ↗	p = 0.009

(Wilcoxon)

	5.9% ↘	p = 0.008

(Wilcoxon)

	11.9% ↘	
[51,60]	p = 0.3

(t‐student)

	–	p = 0.11

(t‐student)

	–	p = 0.01

(t‐student)

	0.9% ↗	p = 0.1

(Wilcoxon)

	–	p = 0.005

(t‐student)

	15.8% ↘	
[61,70]	p = 0.52

(Wilcoxon)

	–	p = 0.06

(t‐student)

	–	p = 0.02

(t‐student)

	0.8% ↗	p = 0.005

(t‐student)

	6.4% ↘	p = 0.03

(Wilcoxon)

	15.2% ↘	
Note: p values for each metric are provided. Highlighted in blue are metrics with p < 0.05. Symbols “↗”, “↘”, “–” indicate the evolution of mean values.

Abbreviations: CNA, cell network atypia; NC, nuclei compactness; NV, nuclei volume; STD_NC, standard deviation of nuclei compactness; STD_NV, standard deviation of nuclei volume.

John Wiley & Sons, Ltd.

NC displayed consistently higher values on the cheekbone of Asian females, with an increase of approximately 1.5% to 3%, while the STD_NC exhibited a consistent lower value of around 9% to 11% (Table 6). The increase in NC was also observed on the mandible for age groups [41‐50], [51‐60], and [61‐70], with increases of 1.4%, 0.9%, and 0.8%, respectively. Except for the age group [51‐60], all the others displayed a significant decrease in STD_NC of about 6%. No significant difference was found on the temple for the STD_NC. The NC was found to be significantly different for age groups [41‐50] and [61‐70].

The CNA was found to be significantly decreased for the temple, cheekbone, and mandible in Asian females (Table 6). For the temple, the decrease ranged from 9% for age group [20‐30] (p = 0.02) to 18.41% for age group [41‐50] (p < 0.001). For the cheekbone, the highest difference was found for age groups [20‐30] (p < 0.001) and [61‐70] (p < 0.001) with approximately a 26% decrease. For the mandible, age groups [20‐30] and [31‐40] did not show significant differences. Age groups [41‐50], [51‐60], and [61‐70] had decreases of 11.9% (p = 0.008), 15.8% (p = 0.005), and 15.2% (p = 0.03), respectively.

4 DISCUSSION

Understanding the differences in skin aging between Caucasian and Asian ethnic groups is an important topic for the cosmetic industry to ensure consumer satisfaction with more customised skincare approaches. The cosmetic industry plays a significant role in improving knowledge of the biological mechanisms of aging through research and innovation. Histological evidence supports a thinning of epidermis and dermis associated with increased cellular heterogeneity (size of basal cells) and lower cellular density for melanocytes, Langerhans cells and fibroblasts. 33 , 34 The flattening of the DEJ is also a reported marker of aging. 35 , 36 Skin biopsies are an invasive medical procedure justified for microscopic examination for diagnostic purposes. However, for the study of healthy skin, non‐invasive techniques in line with the standards of the beauty industry should be preferred. Numerous studies reported in the literature focus on the evolution of clinical signs of aging in terms of appearance and biomechanical properties (elasticity, firmness). 37 The collection of data at the cellular level, in vivo, has been made possible by the development of intracutaneous imaging techniques with sufficient spatial resolution for the observation of microstructures. Therefore, comparative studies highlighting the differences between Caucasian and Asian ethnic groups from histological and structural perspectives remain rare in the literature. While patterns have been established in the apparition of macroscopic features such as faster (i.e., earlier) wrinkling and sagging in Caucasian compared with Asian skin, 27 establishing a link between the visible signs of aging and underlying modifications in skin microstructures would not only help to understand the physiological process involved, but also to identify more specific cellular targets to develop topical products with enhanced anti‐aging efficacy. Reflectance Confocal Microscopy (RCM) has played a key role in promoting optical biopsies for in vivo visualization of the skin. 38 The technique has been widely employed for the diagnosis of skin lesions 39 defining descriptors of skin histology, such as the regularity of honeycomb and pattern of keratinocytes in the epidermal layer, the DEJ organisation or the identification of atypical cells. 38 RCM has also demonstrated its potential to characterise age‐related epidermal and dermal changes in healthy skin 12 , 40 and to further investigate the effects of photoaging and chronoaging on underlying skin microstructures. 41 , 42 Recently, a histopathologic scoring system for skin aging has been proposed by Longo et al. 43 while automated data‐driven quantification is emerging to facilitate the interpretation and grading of skin aging. 44 , 45 Multiphoton Microscopy (MPM) is also a relevant tool to quantify variations in skin layers 10 or pigmentation. 13 The resolution achieved with microscopy techniques applied in vivo allows access to the morphological appearance of keratinocytes 46 and other cell types 47 to decipher the cellular processes of aging. Presently, results reported using LC‐OCT imaging illustrate the level of information that can be achieved by computing cellular metrics in 3D to estimate their volume, shape and resemblance. 11 , 21 Segmentation using AI‐based protocols resulted in a CSD of approximatively 30 000 cells/mm2, that can be then used to conduct statistical analysis to highlight age‐dependent changes within the viable epidermis. This information, coupled with the histological parameters of the skin layers, results in a powerful tool to screen large cohorts of different ethnic groups to seek specific quantitative biomarkers. While CSD displayed a correlation with aging for Caucasian females, this metric was not significantly modified in Asian females (Table 4). Nevertheless, NV and STD_NV appeared to increase with age in both populations. Interestingly, CNA that is a metric encompassing heterogeneity and variability in cell size and shape increased significantly with age.

The examination of aging processes in both Asian and Caucasian populations, with a specific focus on discernible facial signs influencing perceived age, is a recurrent theme in existing literature. Notably, there is well‐documented evidence of aging signs in Asian females. The correlation of these features with the aging process has long been a central topic in skin research, leading to various initiatives aimed at establishing multifactorial scoring systems. In 2002, Guinot et al. 48 made a significant contribution by introducing a Skin Age Score derived from parameters such as comedos, milia, pigmented spots, wrinkles, sagging, and the inability to redden. This score was formulated based on a substantial sample of over 300 Caucasian females in France. Subsequent developments included the SCINEXA model by Vierkötter et al., 49 based on observations from 74 Caucasian females in Germany, and the Age Score by Dicanio et al., 50 derived from a large panel of Caucasian volunteers (n = 600). Additional studies incorporated statistical methods, such as Principal Component Analysis (PCA) and Partial Least Square Regression (PLSR), to identify and select relevant parameters for constructing predictive models, involving 173 Caucasian females 51 and 300 Japanese 52 females, respectively.

However, observations from current studies highlight two main points. Firstly, there is a lack of consistency in studies describing aging scoring on various parameters, including diverse features obtained from facial optical imaging and objective image analysis (sagging, wrinkles, pigmented spots, tone, etc.), biophysical parameters (corneometer, tewameter, sebumeter, etc.), and biomechanical parameters (cutometer, ballistometer). Secondly, there is a noticeable scarcity of studies incorporating multiple ethnicity groups for direct comparison. This disparity in study design persists in recent studies reported by He et al., which focused on 99 Chinese females based on visible, biophysical, and biomechanical parameters, 53 and Pardo et al., which studied Caucasian females based on global wrinkling, pigmented spots, telangiectasia, perceived age, Griffiths grading, actinic keratosis, and keratinocyte cancers. 54 To be noted also that some features used are pathological hence the scoring do not necessarily apply to heathy skin alone.

The need for a comprehensive and inclusive exploration of aging in Caucasian and Asian populations has been emphasized by Robic et al. 37 This involves employing scoring systems or conducting descriptive exploratory studies to gain a better understanding of mechanisms and diverse skin types at the macroscopic level. This imperative extends to studies investigating microscopic modifications in subsurface skin structures related to the aging process. For instance, a recent study by Wang et al. in 2023 examined 189 healthy volunteers aged 22−75 years, categorizing them into young (< 40 years old) and old (≥40 years old) groups. 55 Using two frequencies, 22 and 75 MHz, for HFUS acquisitions, the mean epidermis thickness for the cheek in females was found to be approximately 115 µm. However, first, HFUS resolution was insufficient for accurate measurements, with axial resolutions of ≈60 µm at 22 MHz and ≈20 µm at 75 MHz. Second, the manual segmentation employed most likely resulted in an overestimation of the epidermis thickness due to limitations in image resolution, preventing differentiation between the epidermis and the superficial dermis of the skin. This separation is, however, clearly observed using LC‐OCT imaging, as illustrated in Figure 1. Two studies by Liao et al. 56 and Lin et al. 29 aimed to provide a comparison of aging on the ventral forearm for Asian and Caucasian females, respectively. Harmonic Generation Microscopy was used, a technique with better performance in terms of resolution. The average maximum and minimum thickness of the viable epidermis did not show significant variation with age in the Asian group. LC‐OCT images for the temple, cheekbone, and mandible also did not exhibit significant age‐related variations in Asian females. Notably, an increase in nucleus areas in the forearm was observed, corroborating the size and shape variations seen in LC‐OCT imaging of the face.

The discrepancies in the literature regarding the variation in VE thickness due to aging observed in vivo using non‐invasive imaging were previously discussed in the context of the Caucasian study reported by Bonnier et al. 26 Quite naturally, similar divergences were also noted in studies on Asian volunteers. Chen et al. reported significant differences in epidermis thickness assessed by RCM and MPM, with means of 82 ± 8 and 72 ± 11 µm, respectively, highlighting that the technique employed can influence the results. 57 In their study, authors referred to the epidermis thickness as the thickness of the SC and VE combined. It was observed that the mean obtained with MPM was comparable to the mean obtained with LC‐OCT presently, that is, a full epidermis thickness ranging between 65.6 µm for the age group [51‐60] and 70.4 µm for the age group [41‐50] (Table 1). In contrast, a study by Kutlu Haytoglu et al., which included 120 volunteers from different age groups, found significant variation in supra‐papillary epidermal thickness with aging. 41 However, it is important to note that the borders of the measured thickness were defined as the distance between the first image revealing a honeycomb pattern and the first appearance of dermal tissue inside the circles representing the basal layer of the epidermis, corresponding to the minimum thickness of the epidermis. It's worth mentioning that the mean thickness reported (30–35 µm) differs significantly from the LC‐OCT results and other values describing the VE. In a study conducted by Kawasaki et al., which investigated aging in fifty‐two healthy Japanese subjects (26 males and 26 females) using RCM, notable observations were made. 58 Specifically, it was found that the diameter of granular cells in the skin on the arm and face was significantly larger in the older group (age > 50 years) compared to the younger group (age < 21 years). Furthermore, there was a positive correlation between the diameter of prickle cells in the skin on the face and age. This observation further supports the results found with LC‐OCT regarding variations in volume and compactness for cell nuclei in the VE.

Overall, these findings underscore the imperative for additional research aimed at elucidating and reconciling disparities in our understanding of skin aging across diverse populations. The scientific community necessitates more comparative studies utilising state‐of‐the‐art technologies with optimal resolution to acquire additional valuable and reliable insights. The current study, encompassing over two hundred Caucasian and Asian female volunteers, unequivocally demonstrates the feasibility of quantitatively characterizing signs of aging in the viable epidermis at the cellular level. Consequently, it provides benchmark values for histological and cellular metrics measured in vivo from healthy volunteers, serving as a reference for future exploratory studies on the aging process with inclusive considerations.

5 CONCLUSION

Utilizing 3D LC‐OCT imaging alongside AI algorithms allows for the in vivo quantification of numerous histological and cellular epidermal parameters. This robust technique serves as a powerful tool for investigating age‐related modifications and gaining a nuanced understanding at the micrometric level. Notably, the comparative analysis across different ethnic groups has enabled the identification of both similarities and differences in potential quantitative biomarkers. The observed decrease in the standard deviation of the number of cell layers and cell surface density was specific to Caucasian females. On the other hand, increased mean nuclei volume, heightened standard deviation for nucleus volume, and enhanced cell network atypia emerged as common age‐related features, serving as potential key biomarkers for quantifying facial skin aging in both healthy Caucasian and Asian female volunteers.

This pioneering in vivo study, leveraging 3D LC‐OCT imaging and encompassing two ethnic groups, reinforces the importance of swift and reliable tools for generating extensive datasets. This approach is crucial for deepening our understanding of skin biology and the mechanisms involved in the aging process, all while ensuring inclusivity across diverse populations and skin types.

CONFLICT OF INTEREST STATEMENT

Pedrazzani Mélanie and Lopez Colombe are employees at DAMAE Medical. They contributed to this work under a collaboration with LVMH Recherche to develop tools for 3D LC‐OCT images analysis. All other authors declare no competing interests. Assi Ali, Ralambondrainy Samuel, Grignon Guénolé, Cauchard Jean‐Hubert Korichi Rodolphe and Bonnier Franck are employees at LVMH Recherche. Pignol‐Lavoix Agnes, and Nili Meryem are employees at COMPLIFE, a Contract Research Organization conducting clinical studies. The clinical study was funded by LVMH Recherche.

Supporting information

Supporting Information

ACKNOWLEDGEMENT

None.

DATA AVAILABILITY STATEMENT

All data are available in the text.
==== Refs
REFERENCES

1 Xia X , Chen X , Wu G , et al. Three‐dimensional facial‐image analysis to predict heterogeneity of the human ageing rate and the impact of lifestyle. Nat Metabolism. 2020;2 :946‐957.
2 Luebberding S , Krueger N , Kerscher M . Quantification of age‐related facial wrinkles in men and women using a three‐dimensional fringe projection method and validated assessment scales. Dermatol Surg. 2014;40 :22‐32.24267416
3 Vergilio MM , Vasques LI , Leonardi GR . Characterization of skin aging through high‐frequency ultrasound imaging as a technique for evaluating the effectiveness of anti‐aging products and procedures: a review. Skin Res Technol. 2021;27 :966‐973.33788312
4 Mamalis A , Ho D , Jagdeo J . Optical coherence tomography imaging of normal, chronologically aged, photoaged and photodamaged skin. Dermatol Surg. 2015;41 :993‐1005.26322560
5 Pezzini C , Ciardo S , Guida S , et al. Skin ageing: clinical aspects and in vivo microscopic patterns observed with reflectance confocal microscopy and optical coherence tomography. Exp Dermatol. 2023;32 :348‐358.36394180
6 Vergilio MM , Silva SAM , Jales RM , Leonardi GR . High‐frequency ultrasound as a scientific tool for skin imaging analysis. Exp Dermatol. 2021;30 :897‐910 .33905589
7 Lentsch G , Baugh EG , Lee B , et al. Research techniques made simple: emerging imaging technologies for noninvasive optical biopsy of human skin. J Investig Dermatol. 2022;142 :1243‐1252.e1.35461534
8 Babalola O , Mamalis A , Lev‐Tov H , Jagdeo J . Optical coherence tomography (OCT) of collagen in normal skin and skin fibrosis. Arch Dermatol Res. 2014;306 :1–9.24113905
9 Czajkowska J , Juszczyk J , Bugdol MN , et al. High‐frequency ultrasound in anti‐aging skin therapy monitoring. Sci Rep. 2023;13 :17799.37853086
10 Pena A‐M , Baldeweck T , Decencière E , et al. In vivo multiphoton multiparametric 3D quantification of human skin aging on forearm and face. Sci Rep‐uk. 2022;12 :14863.
11 Chauvel‐Picard J , Bérot V , Tognetti L , et al. Line‐field confocal optical coherence tomography as a tool for three‐dimensional in vivo quantification of healthy epidermis: a pilot study. J Biophotonics. 2022:15 :e202100236.34608756
12 Longo C , Casari A , Beretti F , Cesinaro AM , Pellacani G . Skin aging: in vivo microscopic assessment of epidermal and dermal changes by means of confocal microscopy. J Am Acad Dermatol. 2013;68 :e73‐e82.22000768
13 Pena A‐M , Decencière E , Brizion S , et al. In vivo melanin 3D quantification and z‐epidermal distribution by multiphoton FLIM, phasor and Pseudo‐FLIM analyses. Sci Rep‐uk. 2022;12 :1642.
14 Dubois A . Line‐field confocal optical coherence tomography for high‐resolution noninvasive imaging of skin tumors. J Biomed Opt. 2018;23 :106007‐106007.
15 Dubois A , Levecq O , Azimani H , et al. Line‐field confocal time‐domain optical coherence tomography with dynamic focusing. Opt Express. 2018;26 :33534‐33542.30650800
16 Ogien J , Levecq O , Azimani H , Dubois A . Dual‐mode line‐field confocal optical coherence tomography for ultrahigh‐resolution vertical and horizontal section imaging of human skin in vivo. Biomed Opt Express. 2020;11 :1327.32206413
17 Ogien J , Daures A , Cazalas M , Perrot J‐L , Dubois A . Line‐field confocal optical coherence tomography for three‐dimensional skin imaging. Frontiers Optoelectron. 2020;13 :381‐392.
18 Schuh S , Ruini C , Perwein MKE , et al. Line‐field confocal optical coherence tomography: a new tool for the differentiation between nevi and melanomas? Cancers. 2022;14 :1140.35267448
19 Verzì AE , Broggi G , Caltabiano R , Micali G , Lacarrubba F . Line‐field confocal optical coherence tomography of lentigo maligna with horizontal and vertical histopathologic correlations. J Cutan Pathol. 2023;50 :118‐122.36056910
20 Cinotti E , Tognetti L , Cartocci A , et al. Line‐field confocal optical coherence tomography for actinic keratosis and squamous cell carcinoma: a descriptive study. Clin Exp Dermatol. 2021;46 :1530‐1541.34115900
21 Monnier J , Tognetti L , Miyamoto M , et al. In vivo characterization of healthy human skin with a novel, non‐invasive imaging technique: line‐field confocal optical coherence tomography. J Eur Acad Dermato. 2020;34 :2914‐2921.
22 del Río‐Sancho S , Gallay C , Ventéjou S , Christen‐Zaech S . In vivo evaluation of skin of children with LC‐OCT: An objective assessment. J Eur Acad Dermatol. 2023;37 :1897‐1905. doi:10.1111/jdv.19163
23 Pedrazzani M , Cazalas M , Davis A , et al. Comparison of line‐field confocal optical coherence tomography images with histological sections: validation of a new method for in vivo and non‐invasive quantification of superficial dermis thickness. Skin Res Technol. 2020;26 :398‐404.31799766
24 Fischman S , Suppa M , Cinotti E , et al. Non‐invasive scoring of cellular atypia in keratinocyte cancers in 3D LC‐OCT images using Deep Learning. Sci Rep‐uk. 2022;12 :481.
25 Breugnot J , Rondeau D , Bordes S , et al. Utilizing deep learning for dermal matrix quality assessment on in vivo line‐field confocal optical coherence tomography images. Ski Res Technol. 2022;29 :e13221.
26 Bonnier F , Pedrazzani M , Fischman S , et al. Line‐field confocal optical coherence tomography coupled with artificial intelligence algorithms to identify quantitative biomarkers of facial skin ageing. Sci Rep. 2023;13 :13881.37620374
27 Nouveau‐Richard S , Bouillon C , Humbert P , et al. Skin ageing: a comparison between Chinese and European populations. A pilot study. J Dermatol Sci. 2005;40 :187‐193.16154324
28 Tsukahara K et al. Comparison of age‐related changes in wrinkling and sagging of the skin in Caucasian females and in Japanese females. Int J Cosmet Sci. 2004;26 :314‐314.
29 Lin K , Liao Y , Wei M , Sun C . Comparative analysis of intrinsic skin aging between Caucasian and Asian subjects by slide‐free in vivo harmonic generation microscopy. J Biophotonics. 2020;13 :e201960063.31747129
30 Timár F , Soós G , Szende B , Horváth A . Interdigitation index—a parameter for differentiating between young and older skin specimens: interdigitation index. Skin Res Technol. 2000;6 :17‐20.11428937
31 Weigert M , Schmidt U , Haase R , Sugawara K , Myers G . Star‐convex polyhedra for 3D object detection and segmentation in microscopy. 2020 Ieee Winter Conf Appl Comput Vis Wacv. 2020;00 :3655‐3662.
32 Krishnapuram B . XGBoost: a scalable tree boosting system. Arxiv. 2016;785‐794. doi:10.1145/2939672.2939785
33 Low E , Alimohammadiha G , Smith LA , et al. How good is the evidence that cellular senescence causes skin ageing? Ageing Res Rev. 2021;71 :101456.34487917
34 Russell‐Goldman E , Murphy GF . The Pathobiology of skin aging new insights into an old dilemma. Am J Pathology. 2020;190 :1356‐1369.
35 Vashi NA , Maymone MB de C , Kundu RV . Aging differences in ethnic skin. J Clin Aesthetic Dermatology. 2016;9 :31‐38.
36 Roig‐Rosello E , Rousselle P . The human epidermal basement membrane: a shaped and cell instructive platform that aging slowly alters. Biomol. 2020;10 :1607.
37 Robic J , Nkengne A , Lua BL , et al. Modeling of global skin aging indices among Caucasian and Asian women. Ski Res Technol. 2023;29 :e13222.
38 Shahriari N , Grant‐Kels JM , Rabinovitz H , Oliviero M , Scope A . Reflectance confocal microscopy Principles, basic terminology, clinical indications, limitations, and practical considerations. J Am Acad Dermatol. 2021;84 :1‐14.32553679
39 Guida S , Pellacani G , Ciardo S , Longo C . Reflectance confocal microscopy of aging skin and skin cancer. Dermatology Pract Concept. 2021;11 :e2021068.
40 Cinotti E , Bovi C , Tonini G , et al. Structural skin changes in elderly people investigated by reflectance confocal microscopy. J Eur Acad Dermatol. 2020;34 :2652‐2658.
41 Haytoglu NSK . Assessment of skin photoaging with reflectance confocal microscopy. Ski Res Technol. 2014;20 :363‐372.
42 Wurm EMT , Longo C , Curchin C , et al. In vivo assessment of chronological ageing and photoageing in forearm skin using reflectance confocal microscopy. Brit J Dermatol. 2012;167 :270‐279.22428802
43 Longo C , Simonazzi S , Mazzaglia G , et al. Proposal for an in vivo histopathologic scoring system for skin aging by means of confocal microscopy. Skin Res Technol. 2013;19 :e167‐e173.22672873
44 Lboukili I , Stamatas G , Descombes X . Automatic granular and spinous epidermal cell identification and analysis on in vivo reflectance confocal microscopy images using cell morphological features. J Biomed Opt. 2023;28 :046003‐046003.37038547
45 Hames SC , Bradley AP , Ardigo M , Soyer HP , Prow TW . Towards data‐driven quantification of skin ageing using reflectance confocal microscopy. Int J Cosmetic Sci. 2021;43 :466‐473.
46 Koehler MJ , Zimmermann S , Springer S , et al. Keratinocyte morphology of human skin evaluated by in vivo multiphoton laser tomography. Skin Res Technol. 2011;17 :479‐486.21453411
47 Kröger M , Scheffel J , Nikolaev VV , et al. In vivo non‐invasive staining‐free visualization of dermal mast cells in healthy, allergy and mastocytosis humans using two‐photon fluorescence lifetime imaging. Sci Rep‐uk. 2020;10 :14930.
48 Guinot C , Latreille J , Mauger E , et al. Relative contribution of intrinsic vs extrinsic factors to skin aging as determined by a validated skin age score. Arch Dermatol 2002;138 :1454‐1460.12437451
49 Vierkötter A , Ranft U , Krämer U , et al. The SCINEXA: a novel, validated score to simultaneously assess and differentiate between intrinsic and extrinsic skin ageing. J Dermatol Sci. 2009;53 :207‐211.19059763
50 Dicanio D , Sparacio R , Declercq L , et al. Calculation of apparent age by linear combination of facial skin parameters: a predictive tool to evaluate the efficacy of cosmetic treatments and to assess the predisposition to accelerated aging. Biogerontology. 2009;10 :757.19333774
51 Nkengne A , Roure R , Rossi AB , Bertin C . The skin aging index: a new approach for documenting anti‐aging products or procedures. Ski Res Technol. 2013;19 :291‐298.
52 Miyamoto K , Inoue Y , Hsueh K , et al. Characterization of comprehensive appearances of skin ageing: an 11‐year longitudinal study on facial skin ageing in Japanese females at Akita. J Dermatol Sci. 2011;64 :229‐236.22015051
53 He Y , et al. Establishment of a PCA model for skin health evaluation. Biotechnol Biotechnol Equip. 2018;32 :1060‐1064.
54 Pardo LM , Hamer MA , Liu F , et al. Principal component analysis of seven skin‐ageing features identifies three main types of skin ageing. Br J Dermatol. 2020;182 :1379‐1387.31519034
55 Wang S , Yu RX , Fan W , et al. Detection of skin thickness and density in healthy Chinese people by using high‐frequency ultrasound. Ski Res Technol. 2023;29 :e13219.
56 Liao Y‐H , Chen SY , Chou SY , et al. Determination of chronological aging parameters in epidermal keratinocytes by in vivo harmonic generation microscopy. Biomed Opt Express. 2013;4 :77‐88.23304649
57 Chen K , Wang Z , Han Y , Cui Y . In vivo detection of healthy skin: comparison of multiphoton microscopy and reflectance confocal microscopy. Ski Res Technol. 2023;29 :e13340.
58 Kawasaki K , Yamanishi K , Yamada H . Age‐related morphometric changes of inner structures of the skin assessed by in vivo reflectance confocal microscopy. Int J Dermatol. 2015;54 :295‐301.25267556
