Establishment of visual assessment for the severity of dark circles in Chinese Han women

Abstract Objective To establish a visual assessment scale for the severity of dark circles' pigmented and vascular type in Chinese women. Materials and method A total of 269 healthy Chinese women from Shanghai with different degrees of dark circles', both pigmented and vascular types, were evaluated by visual assessment. Photographs of their dark circles were analyzed by image analysis. Results The visual assessment evaluation on classification and severity showed a favorable agreement between the successive measured results. Significant differences from very slight to severe dark circles for pigmented type and vascular type were observed. The severity level by visual assessment was significantly positively correlated with ΔE values while negatively correlated with ΔL values (P < .01) in both pigmented and vascular types. Besides, Δa values of vascular type were significantly positively correlated with the ΔE values for dark circles' vascular type. Values between ΔE and ΔL also showed a significant negative correlation (P < .01). Conclusion The five‐point visual assessment scale for dark circles of vascular and pigmented types was verified and proved to have good repeatability. The image analysis's objective result proved significant and consistent with the visual assessment and color parameters. This scale could be a useful and effective tool in diagnosing dark circles' severity.

type is the pigmented type. The leading cause for its formation is excessive pigmentation, which is seen in dermal melanocytosis, allergic, and contact dermatitis caused by postinflammatory pigmentation. 6,7 Dark circle is a significant cosmetic concern. Patients, mostly women, are found to be commonly affected. Studies on dark circles show that a significant number of these patients were young adults between 20 and 30 years of age. 4 They also complained of dissatisfaction with their facial appearance, fatigue, and insomnia. Lack of sleep was reported to be the major factor to dark circle in Asian population. 8 Besides, stress or anxious, emotional liability, and aging can play a significant role in developing dark circles. 8,9 Although skin conditions under the eyes would not threaten physical health, dark circles influence facial appearance and make people look tired, sad, or lethargic, causing negative emotions. Hence, people invariably tend to seek treatment for this skin problem. More eye skincare products in the market to improve this problem.
Notably, the dark circle has gained more attention from consumers, especially women. Due to the individual basis and numerous causes, it is hard to define and classify dark circles. Previous studies have set up objective methods to measure dark circles according to their characteristics and causes. 5,10 Ultrasonography, wood's lamp, and image analysis were primarily applied for classification. Some studies have reported evaluating the degree of dark circles in general types. 10 However, the assessment for the severity of an individual type of dark circle is yet to be reported. What's more, there is still lack of literature that focus on Chinese Han women although the problem of dark circle may be of greater concern in this population. It is reported the distribution of dark circle types varied among countries. Among Chinese population, vascular type with lighter skin (type I-IV) is more frequent compared with other Asian population. 4,8 Pigmented type is another common type owing to excessive pigmentation under eyes in Asian population. It could be seen in conditions including dermal melanocytosis and postinflammatory hyperpigmentation. 6 It presents irregular patches of brownish or gray pigmentation on the eyelids as a result of color transmission of gray or blue-gray melanocytoses through the dermis. 4,6 Structural type of dark circle especially with tear trough is prevalent throughout 3 races in Asia, including Chinese, Malays, and Indians but less occurred among Chinese population. 8 It is caused by shadowing which could be either tear trough or caused by skin laxity, puffy, or the overlying pseudoherniation of the infraorbital fat. 1,11 In this case, different angles of light could affect the appearance of dark circle. Thus, this study aimed to firstly establish a visual assessment scale to individually evaluate the severity of both pigmented and vascular types in Chinese women.
The established visual assessment scale will provide a supplement method for dark circles, which can also be applied in clinical research. The first step is to classify different types of dark circles.
Notably, the primary symptom of dark circles is the change of color characteristic under the eyes. The dark circles were classified according to their color appearance by visual assessment to exclude an individual basis. The structural type of dark circles is excluded due to its structural shadows caused by skin contours. After classification, the severity level was scored based on the color difference between native skin tone and the periorbital skin. Photographs of dark circles were further analyzed by image analysis to verify the reliability between the color parameters obtained by image analysis and severity level by visual assessment.

| Subjects population
Two hundred sixty-nine healthy Chinese women from Shanghai with different degrees of pigmented type or vascular type of dark circles were enrolled. The age ranged from 18-60 years old. The dark circles were classified according to appearance characteristics through visual assessment. This study was approved by Shanghai institutional research ethics committee (Application Number: SECCR/2019-38-01) and all patients signed consent forms.
Before the evaluation, subjects were asked to wash their face using a standard cleanser and stay in a constant temperature-controlled room for at least 30 minutes (20-24°C, 40%-60% room humidity).
The photograph of the subjects' front face was collected by VISIA-CR (Canfield, Fairfield) under a standard light source (Mode 2). These photographs would be evaluated in the subsequent assessment.

| Visual assessment
Experienced dermatologists evaluated the visual assessment of dark circles. Firstly, dark circles were classified into a pigmented or vascular type according to their clinical color appearance and causes. 5,7,12 The vascular type of dark circle appears partially cyan, mainly located on the lower eyelid. The color will be lightened after pulling down the lower eyelid. This type has relatively thin eyelid skin with poor blood circulation, leading to hypoxia and microvascular blood veins' detention. It appears tawny and primarily located on the lower eyelid but, with the larger involved area for pigmented type. The

| Image analysis
Images of dark circles from Visia-CR were evaluated by the image analysis program (Image-Pro Plus). The colors of dark circles (under eyes) and native skin tone (cheek) were measured. The results were reported as the L, a, and b color space. The color difference (ΔE) between the dark circles and the native skin tone was calculated using the equation below: where ΔL, Δa, and Δb were the differences in the values of lightness (L), redness-to-greenness (a), and yellowness-to-blueness (b) between the dark circles and native skin tone.

| Statistical analysis
The differences among groups of different degrees in each group were analyzed using one-way ANOVA. Kappa test and intraclass correlation coefficient were used to evaluate test-retest reliability.
Spearman's P-value was used to calculate the correlation between image analysis and visual assessment. Statistical analysis was performed using SPSS software (Version 24, IBM). A significant difference was measured at P < .05.

| Conformity assessment
Repeated evaluation of 100 images of the dark circle was conducted on the same computer screen to verify the reliability of visual assessment for dark circle images. The result showed that visual assessment had good repeatability of the classification and severity score. For the reliability of classification, the correlation coefficient (r) is .762 (P < .01), and for the severity score, r = .823 (P < .01).

| Severity of dark circle evaluation by visual assessment and image analysis
The severity of dark circles was evaluated by a newly developed visual assessment scale and image analysis. Firstly, 269 subjects with different dark circles were classified into two types by the dermatologist according to the dark circles' definition and color appearance.
The proportion of pigmented type is 45%, and vascular type is 55%.
By using the newly developed scale for dark circle severity with the range of 0-4 score, it is found that there were significant differences from very slight to severe level of dark circles. In the result of vascular type, the most proportion was 52.4% from trim (slight) level (score 1.5 and 2), while others were 25.9% (very slight level; score 0.5 and 1), 20.8% (moderate level; score 2.5 and 3), and 1.0% (severe level; score 3.5). At the level of a severe degree, the highest score was 3.5. Notably, none of these subjects scored 4 ( Table 1).
The result of image analysis showed that the color difference values (ΔE) increased from 8.06 to 16.14, while the difference values of lightness (ΔL) decreased from −7.18 to 15.49 as the severity scores increased from 0.5 to 3.5 ( Figure 3).
Furthermore, as shown in Figure 4A and 11.5% from severe level (score 3.5 and 4) ( Table 2). Image analysis showed that as the severity scores increased from 0.5 to 4, ΔE values significantly increased from 10.42 to 17.92, while ΔL values decreased from −9.91 to −17.18 ( Figure 5). Interestingly, Figure 6A showed that the severity level of pigmented type by visual evaluation is significantly correlated with the ΔE and ΔL values (P < .01).

| D ISCUSS I ON
Despite the significant prevalence of dark circles, it remained an unsolved and complex problem to diagnose this clinical manifestation.
Different types of dark circles have their unique characteristics. 4 We proposed that it is necessary to evaluate each type with different as- can than dermis at other sites, 13 presented a thinner eyelid suggested by ultrasound images analysis. 14 In this case, it was proposed that the skin color of lower eyelid appeared much redder in the vascular type of dark circle due to the clogged cutaneous blood and thinner eyelids as the severity level increased.
However, the proposed scale still has some weaknesses. Firstly, some fluctuations were observed in the color parameters (ΔE and ΔL), especially at the half-point level. Table 1 and Table 2   Note: ΔE, the color difference between dark circles and native skin tone; ΔL, the difference of lightness between dark circles and native skin tone; Δa, difference of redness-to-greenness between dark circles and native skin tone; Δb, the difference of yellowness-to-blueness between dark circles and native skin tone.
TA B L E 1 Color parameters on the vascular type of dark circles among different severity levels (Mean ± SD)

F I G U R E 3
Changes in visual assessment and image analysis of vascular type of dark circles. ΔE, the color difference between dark circles and native skin tone; ΔL, the difference of lightness between dark circles and native skin tone; Δa, difference of redness-togreenness between dark circles and native skin tone  Note: ΔE, the color difference between dark circles and native skin tone; ΔL, the difference of lightness between dark circles and native skin tone; Δa, difference of redness-to-greenness between dark circles and native skin tone; Δb, the difference of yellowness-to-blueness between dark circles and native skin tone.

TA B L E 2
Color parameters on the pigmented type of dark circles among different severity levels (Mean ± SD) F I G U R E 5 Changes in visual assessment and image analysis of the pigmented type of dark circles. ΔE, the color difference between dark circles and native skin tone; ΔL, the difference of lightness between dark circles and native skin tone

ACK N OWLED G EM ENTS
We thank Hairong Liang for the assistance with dark circle evaluation.

CO N FLI C T O F I NTE R E S T
This study is conducted based on clinical tests and has got permis-