Dynamic visualization of ultraviolet dose on skin with sunscreen applied using minimum erythema dose

Abstract

Background

Visualizing the ultraviolet (UV) dose on skin serve as an intuitive approach to ensure appropriate sunscreen usage and reduce the risk of erythema. UV dose is determined by a number of external factors, such as properties of sunscreens, weather, and type of outdoor activity. We propose a framework for visualizing UV doses that considers various external factors.

Materials and methods

First, the skin of a three‐dimensional human model was represented using triangular meshes, and various static postures and dynamic motions were simulated to express outdoor activities. Then, we evaluated the persistency and insufficiency properties of sunscreen, which are time dependent and directly affect the effectiveness of the sunscreen skin protection factor (SPF) during UV exposure. Finally, to calculate the UV dose in real time, we tracked the trajectory of the sun and motion of the skin while considering the time‐dependent properties of sunscreen.

Results

An S/W system was implemented based on the proposed framework to visualize the distribution of UV doses through dynamic color changes in exposed skin areas. The color types include true colors, which represent the minimum erythema dose (MED), and pseudo colors representing states before 1 MED is reached. We devised various examples to discuss the usability of the proposed framework.

Conclusion

The system conveniently displays the MED according to an individual's skin phototype. When the properties of a wide range of commercial sunscreens are added to the system database, it is expected that the rate of appropriate sunscreen usage by customers will increase.

1. INTRODUCTION

UV radiation (UVR) is one of the main causes of skin aging, pigmentation, skin cancer, and other skin diseases. ¹ , ² , ³ Numerous studies have proved that the use of sunscreen is a safe and effective method to prevent skin cancer. ⁴ , ⁵ , ⁶ Van der Pols et al. ⁷ demonstrated that regular sunscreen application can reduce the incidence of squamous cell carcinomas by 40%. Therefore, prediction of UV doses on the skin and proper sunscreen usage could be essential components of skincare to avoid erythema during outdoor activities.

The wavelength band of UVR is 100−400 nm, and this band is divided into UVA (315−400 nm), UVB (280−315 nm), and UVC (100−280 nm). The UV component of terrestrial radiation from the midday sun contains approximately 95% UVA and 5% UVB. ³ , ⁸ , ⁹ , ¹⁰ UVC and most UVB are removed from extraterrestrial radiation by the stratospheric ozone layer. ¹¹ Among UVR, UVB is considered the leading cause of erythema. ³ , ⁹ , ¹² UVA, too, causes erythema, albeit at significantly higher dosages than UVB. The amount of UVB that starts to induce erythema is defined as minimum erythema dose (MED), which varies depending on the Fitzpatrick skin phototype (Table 1). ³ Therefore, by predicting the UVB dose based on an individual's skin type, one can effectively prevent the risk of sunburn during outdoor activities. In this study, we assume that the effect of UVA on erythema is negligible.

TABLE 1.

Fitzpatrick skin phototype scales and UVB‐MEDs. ³

Fitzpatrick phototype	Cutaneous response to UV	UVB‐MED (mJ/cm2)
I	Always burns Never tans	15–30
II	Burns easily Tans minimally	25–40
III	Burns moderately Average tanning ability	30–50
IV	Burns minimally Tans easily	40–60
V	Rarely burns Tans easily and substantially	60–90
VI	Almost never burns Tans readily and profusely	90–150

To compute amount of UVR, we first define UV irradiance as the amount of solar power incident on a unit area (W/cm² or mJ/s/cm²), UV irradiation as the amount of solar energy incident on a unit area (mJ/cm²), and UV dose as the energy actually absorbed by a unit area of skin (mJ/cm²). Assuming that the amount of UVB is 1/20 (5%) of UVR, ⁸ , ⁹ , ¹⁰ the UVB dose to skin over an exposure time of t_s to t_e can be computed as follows:

$$\mathrm{UV}{B}_d=\frac{1}{20}{\int }_{t_s}^{t_e}\frac{U{V}_i(t)}{\mathrm{SPF}(t)}\mathrm{dt}(\mathrm{mJ}/{\mathrm{cm}}^2),$$ (1)

where $U{V}_i(t)$ (mJ/s/cm²) is the UV irradiance incident on a unit area of skin, and $SPF(t)$ is the sun protection factor. Because both UV_i and skin protection factor (SPF) change with exposure time, many studies have been conducted to accurately estimate their changes.

1.1. Estimation of UV irradiation on skin

Most early studies used dosimeters to measure the UV exposure of an individual in a given environment. Kimlin et al. ¹³ presented a method to estimate the UV irradiation on a human face by using dosimeters set on manikins and drew the exposure level in the form of a series of contour plots over the image of a human face. Downs and Parisi ¹⁴ extended this approach by using a geometric model of a human face to visualize the level of UV exposure with several colors. Hoeppe et al. ¹⁵ used pseudo colors to visualize the UV irradiation measured over the entire body of a three‐dimensional (3D) model in different postures. However, relying solely on measurements is time‐consuming, and the approach cannot account for a wide range of environmental changes.

Several recent studies have focused on UV exposure calculations instead of measurements. Vernez et al. ¹⁶ proposed a numerical model to predict and display UV exposure on a 3D human model in static postures by using pseudo colors. To validate the accuracy of their numerical model, they compared the predicted data with measured values. Backes et al. applied Vernez et al.’s simulation technique to evaluate the effectiveness of sun protection on a face by using different hat styles and sunglasses. ¹⁷ , ¹⁸ Similar to the approach of Vernez et al., ¹⁶ Salvadori et al. ¹⁹ proposed an algorithm to predict UV irradiation on sloped surfaces representing different parts of the human body. However, all of the early and recent studies have not considered dynamic motions of the human body and the sun and have not addressed sunscreen application.

1.2. Persistency of sunscreens

The quantity of sunscreen applied to skin does not persist over time because routine activities reduce the residual quantity. ²⁰ , ²¹ Therefore, even if a sufficient amount of 2 mg/cm² is applied initially, the US FDA(Food and Drug Administration) recommends that sunscreen should be reapplied every 2 h. ²² , ²³ , ²⁴ The persistency of sunscreen directly influences the effectiveness of SPF, and therefore, it must be considered in Equation (1) to increase the accuracy of UV dose computation.

Rungananchai et al. ²⁵ tested the persistency of sunscreen by measuring the remaining quantity of sunscreen on different parts of the facial area at intervals of 2 h. Several studies ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ have tested various commercial products to derive a relationship between the quantity and effectiveness of SPF. These studies have reported that the effectiveness of SPF increases linearly or exponentially with its quantity. In addition, even for different products with the same SPF value, it was found that the effectiveness differed depending on the ingredients. However, none of the previous studies have addressed changes in the effectiveness of SPF over time after sunscreen application, as required in Equation (1).

In this study, we propose a framework for dynamically visualizing the UVB dose on the skin while considering various personal and environmental factors related to outdoor activities (Figure 1). The framework consists of three main parts:

1. User inputs: These include users’ choices of environmental and personal factors such as location, and date and time of outdoor activities, and sunscreen products.

2. Weather and sunscreen databases: These contain archived global UV irradiance data provided by local meteorological authorities, as well as the properties of commercial sunscreen products, such as changes in quantity and SPF effectiveness over time after sunscreen application.

3. UV dose simulation and visualization: This iterates the $U{V}_i(t)$ and $SPF(t)$ computations based on user input data and the aforementioned databases and dynamically displays UVB dose distribution by using color spectra.

FIGURE 1.

Schematic diagram representing the framework

To formulate the changes in SPF effectiveness over time, we performed experiments involving organic and inorganic sunscreens to determine their residuals over time and combined the results with those of previous studies. ²⁶ , ²⁸ To visualize changes in the UVB dose, entire meshes representing the human skin were rendered in real time with the corresponding true colors related to UVB‐MED or pseudo colors.

2. MATERIALS AND METHODS

2.1. 3D human models

To display skin color changes induced by UVB doses, 3D human body models of a man and woman were created. For more detailed and accurate visualization, the skin of each model was represented using approximately 22,000 triangular meshes. Each mesh contained the data of vertices, normal vectors, and color.

To simulate various postures or model motions, a hierarchical set of skeletons was added to the model by following a rigging process. Four static body postures (supine, standing, sitting, and driving) and two dynamic motions (walking and running) were implemented. Then, we implemented a skinning process that combined the skeleton and the skin, which allowed the skin to deform more realistically and smoothly during motion simulation of the human models. In this study, the 3D models were created using Zbrush, a digital sculpting tool for 3D modeling, texturing, and painting. The processes of rigging and skinning were implemented using Blender, an open‐source software application.

2.2. UV irradiance on meshes

The amount of UVR that reaches the skin $U{V}_i(t)$ is affected by multiple environmental factors, such as geographic location, weather conditions, date and time, and skin orientation with respect to the sun. These orientations change dynamically during outdoor activities and significantly affect the UV dosage. Therefore, in this study, to improve calculation accuracy, we tracked the positions of the sun and skin in real time and included them in the $U{V}_i(t)$ calculation.

In principle, the global irradiance reaching the skin is the sum of three components: reflected, direct, and diffuse. ³¹ , ³²

$$U{V}_i\left(t\right)=U{V}_{\mathrm{ref}}\left(t\right)+U{V}_{\mathrm{dir}}\left(t\right)+U{V}_{\mathrm{dif}}\left(t\right)\left(\mathrm{mJ}/s/{\mathrm{cm}}^2\right)$$ (2)

The formulas for computing each of these terms can be found in the literature. ³¹ , ³² We express them by using the angular relationship among the sun, skin (a tilted mesh), and the horizontal plane illustrated in Figure 2, as follows:

$$U{V}_{\mathrm{ref}}\left(t\right)=\frac{1}{2}\left(1-\cos \beta \right)R\left(t\right)\left(0\le \beta \le \pi \right)$$ (3a)

$$U{V}_{\mathrm{dir}}\left(t\right)=\frac{\cos {\theta }_i}{\cos \left(90-\alpha \right)}B\left(t\right)\left(0\le {\theta }_i\le \frac{\pi }{2}\right)$$ (3b)

$$U{V}_{\mathrm{dif}}\left(t\right)=\frac{1}{2}\left(1+\cos \beta \right)D\left(t\right),$$ (3c)

where R(t), B(t), and D(t) are the reflected, direct, and diffuse hourly irradiance in an average day of each month. ³¹ , ³² Here, R(t) is mainly influenced by the type of surrounding ground. It is computed using the relationship R(t) =  $\rho H(t),$where ρ is the reflectivity of the surrounding ground (grassland: 0.22, asphalt road: 0.07, sea water: 0.08, beach sand: 0.17, snow: 0.69), ³³ and H(t) is the total irradiance incident on the horizontal plane. In general, H(t) data can be obtained from local meteorological authorities. In our framework, the $H(t)$ data of major cities worldwide are included in the database for easy access. B(t) is determined using the relationship B(t) = H(t)‐D(t), in which D(t) varies with weather conditions. ³¹ , ³² We have considered three types of weather conditions in our framework: clear, cloudy, and overcast. The angles α, β, and θ _i defined in Figure 2 change with time, and they represent skin orientation with respect to the sun. In this study, these angles were calculated for each second by using celestial relationships of the solar system. For brevity, the calculation process is omitted from this paper.

FIGURE 2.

Global irradiance on a tilted mesh representing skin

2.3. Persistency, insufficiency, and SPF(t)

In general, consumers do not fully utilize the effectiveness of SPF labeled on a product for two main reasons:

1. Persistency: The quantity of applied sunscreen decreases over time because of wear. ²⁵ , ³⁴

2. Insufficiency: Most consumers apply typically between 0.5 and 1.5 mg/cm², ²⁷ , ²⁹ which is less than 2 mg/cm², the effective sunscreen dosage recommended by the US FDA.

To formulate the $SPF(t)$ of a given sunscreen, persistency, and insufficiency must be considered together because a lack of persistency leads to insufficiency. During experimentation, it may be necessary to simultaneously measure the residual quantity and calculate the corresponding SPF effect. However, there is a time lag of several hours between quantity measurement and SPF calculation using MED, which may require intensive efforts. Instead, to formulate $SPF(t$), we propose the following three steps: (1) formulate Q(t), which is defined as a persistency function representing changes in the residual quantity over time; (2) formulate $SPF(q)$, which is defined as an insufficiency function representing changes in SPF with the amount of sunscreen applied q; and (3) substitute q from $SPF(q)$ in Q(t) to derive $SPF(t$). Notably, Q(t) has many different forms depending on the sunscreen application conditions. ²⁵ This is true for$SPF(t)$ as well.

• Step (1): First, we conducted an experiment to formulate Q(t) for two different types of sunscreens. We selected two of the most popular commercial products in Korea: Product A, an inorganic sunscreen with SPF35 and Product B, an organic sunscreen with SPF43. To reflect the reality of insufficiency, our volunteers applied only two‐thirds of 2.0 mg/cm² (1.33 mg/cm²) ²⁷ , ³⁵ on their facial skin and continued their routine indoor and outdoor activities. Then, fluorescence intensity images captured using a facial imaging system (VISIA‐CR, Canfield Scientific, Fairfield, NJ, USA) were analyzed to determine the residual quantity of sunscreen. Each dot in Figure 3 represents the mean value of the residual quantity of each product measured at intervals of 2 h. Finally, through regression analysis, the persistency function Q(t) was formulated in the form of quadratic equations:

  $$\begin{array}{ccc}\hfill \mathrm{Product}A:& & Q{\left(t\right)}_{\mathrm{inor}35}=6\times {10}^{-10}{t}^2-5\times {10}^{-5}t\hfill \\ & & +1.33\left(\mathrm{mg}/{\mathrm{cm}}^2\right)\hfill \end{array}$$ (4a)

  $$\begin{array}{ccc}\hfill \mathrm{Product}B:& & Q{\left(t\right)}_{\mathrm{or}43}=7\times {10}^{-10}{t}^2-6\times {10}^{-5}t\hfill \\ & & +1.33\left(\mathrm{mg}/{\mathrm{cm}}^2\right)\hfill \end{array}$$ (4b)
• Step (2): To formulate the insufficiency function $SPF(q)$, we reviewed the results of previous studies ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ and deduced that because $SPF(q)$ is a unique characteristic of sunscreen, the functions defined in the previous studies have common forms depending on the type of sunscreen: the $SPF(q)$ of the products with SPF values lower than 30 were linear functions ²⁶ , ²⁷ , and the $SPF(q)$ of the products with SPF values higher than 30 were exponential functions. ²⁰ , ²² , ²³ In the present study, instead of conducting another experiment to evaluate $SPF(q)$, we used the results of previous studies to simplify S/W development and enhance the validity of our framework. Among them, we selected the function of an inorganic sunscreen with SPF 30 ²⁸ and that of an organic sunscreen with SPF 55 ²⁶

  $$\mathrm{SPF}{\left(q\right)}_{\mathrm{inor}\mathit{30}}=e^{1.509+1.0329\times q},$$ (5a)

  $$\mathrm{SPF}{\left(q\right)}_{\mathrm{or}\mathit{55}}=2.8585e^{1.4\times q},$$ (5b)

  because the SPF values and ingredients of these sunscreens are similar to those of products A and B. Equations (5) were derived by conducting an in vivo experiment with four different quantities of sunscreen (0.5, 1.0, 1.5, 2.0), and the SPF values were calculated as SPF = (MED with sunscreen/MED without sunscreen). ²⁶ , ²⁸
• Step (3): We transformed Equation (5) into a function of time by substituting Q(t) for q and obtaining $SPF(q)$:

  $$\mathrm{SPF}{\left(t\right)}_{\mathrm{inor}\mathit{30}}=e^{1.509+1.0329\times Q{\left(t\right)}_{\mathrm{inor}35}}$$ (6a)

  $$\mathrm{SPF}{(t)}_{\mathrm{or}\mathit{55}}=2.8585e^{Q{\left(t\right)}_{\mathrm{or}43}}$$ (6b)

where Figure 4 illustrates the SPF curves of these functions.

FIGURE 3.

Persistency functions Q(t) for products A and B. R ² is the correlation coefficients

FIGURE 4.

Skin protection factor (SPF)(t) for product A and product B

Now, by rearranging Equation (1) based on Equations (3) and (6), we can finally formulate the total UVB dose accumulated on each mesh until time t by summing the UVB dose calculated at each time step (Δt) from the start of exposure, as follows:

$$\begin{array}{ccc}& & \mathrm{UV}{B}_d\left(t\right)=\frac{1}{20}\sum _{t=t_s}^t\frac{1}{\mathrm{SPF}\left(t\right)}\hfill \\ & & \left[\frac{1}{2}\left(1-\cos \beta \right)\rho H\left(t\right)+\frac{\cos {\theta }_i}{\cos {\theta }_z}B\left(t\right)+\frac{1}{2}\left(1+\cos \beta \right)D\left(t\right)\right]\mathrm{\Delta }t\left(\frac{\mathrm{mJ}}{{\mathrm{cm}}^2}\right),\hfill \end{array}$$ (7)

where  Δ t = 1 (s) for our visualization, which will be discussed in detail in the following section.

3. RESULTS AND DISCUSSION

Figure 5 illustrates the user interface of the software (S/W) system based on the proposed framework and implemented in the Unity platform. The upper part of the screen contains the menus for selecting personal and environmental factors. The lower part of the screen displays the distribution of UVB doses on a human model by using dedicated colors. Five examples were provided herein to exemplify the results of the proposed methods. In all of these examples, it was assumed that only two‐thirds of the effective dose of 2.0 mg/cm² (1.33 mg/cm²) was applied initially.

FIGURE 5.

User interface of the S/W system

Example 1

Color mapping

Figure 6 visualizes the distribution of UVB doses at 2h intervals when a man with skin phototype II stands in a static posture facing south. The spectrum of red, which representing the UVB dose, is expressed in terms of the MED of the Fitzpatrick skin phototype (Table 1). ³ Hence, for the same UVB dose, the intensity of red color may appear differently in the following figures depending on the skin phototype. The intensity of red color in Figure 6 indicates the degrees of UVB‐MED from 1 to 5, and this is true for all other relevant figures in the paper. As a precaution, the lower MED value of each phototype in Table 1 is used in the pictures. For example, the UVB value corresponding to 1 MED in Figure 6 is 25 mJ/cm², which is for phototype II. The lettered arrows in the MED spectrum indicate the maximum UV dose for each case. Note that the color distribution in Figure 6 is asymmetric because UVR for 3 h in the afternoon is stronger than that in the morning.

FIGURE 6.

Phototype II, SPF 30 inorganic, Sydney, clear, 9:00−15:00, October 31, no motion, facing south

Example 2

UV reflectivity of ground

Figure 7 demonstrates the effects of two types of surrounding grounds: sand and grassland. The intensity of red color in Figure 7A is stronger than that in Figure 7B because the UV reflectivity of sand is higher than that of grassland. Compared to Figure 6, UVB doses in Figure 7 are more evenly distributed owing to motion of the skin during exposure.

FIGURE 7.

Phototype III, SPF 30 inorganic, Seoul, clear, 10:00−14:00, June 22. (A) Sand. (B) Grassland

Example 3

Skin phototypes and outdoor activities

Figure 8 visualizes the distribution of UVB doses on two men with two different skin phototypes: I and III. The intensity of red color indicates that the man with phototype I reaches 2 MED within 6 h of either standing or running in New York City on a clear afternoon in August (Figure 8A,B). By contrast, the man with phototype III reaches only 1 MED under the same conditions, which is 50% lower (Figure 8C,D). Note that the UVB doses in Figure 8A,C are concentrated on certain parts of the skin because the man is not moving. By contrast, the UVB doses in Figure 8B,D are evenly distributed because the man is moving.

FIGURE 8.

(A and B): phototype I, (C and D): phototype III, SPF 30 inorganic, New York City, clear, 12:00−18:00, August 22

Example 4

Reapplication and weather

Figure 9 shows the case of sunscreen reapplication. Compared to the case of no reapplication (Figure 9A), single reapplication within 8 h significantly reduces UVB doses, but there are signs of erythema. By contrast, in the case of reapplication after 2 h (Figure 9C), there are no signs of erythema, which implies the necessity of reapplication at intervals of 2 h. Upon reapplication, theSPF(t) curve was assumed to behave in the same manner as that after the initial application. Figure 9D demonstrates the importance of using sunscreen even on cloudy days.

FIGURE 9.

Phototype II, Sydney, 10:00−18:00, October 31, asphalt road, facing south, (A–C) clear day, (D) cloudy day. (A) No application, (B) skin protection factor (SPF) 50 organic w/o reapplication, (C) reapplication after 2 h. (D) No application

Example 5

I ndoor activity

We applied the proposed method to an indoor activity, driving. It was assumed that sunscreen was not applied, and the sun was incident only from the front and side windows of the car. As expected, for phototype III, erythema was not reached even after 6 h of driving in the afternoon. Therefore, UVB doses were expressed using intensities of pseudo color green instead of red (Figure 10).

FIGURE 10.

No sunscreen, phototype III, Seoul, clear, 12:00−18:00, June 22

4. CONCLUSION

In this paper, we proposed a framework for dynamically calculating and visualizing UVB doses on the skin. The framework consists of three main parts: The first part is the user interface, where a user can enter personal and external factors. The second part is the database that stores global UV irradiance data for each major city and property information of commercial sunscreens such as $SPF(t)$. The third part is the engine that computes the UVB dose in real time. It calculated the UV irradiance by tracking the motions of people and the sun for a given location and period of time. Moreover, changes in SPF are included in the calculation. Finally, the UVB dose distribution on the skin of a human model at each time step is dynamically displayed by using appropriate colors.

The persistency function Q(t) can vary depending on application conditions, such as type of outdoor activity, even for the same product. If Q(t) values for various application conditions are available, the accuracy of UVB dose calculation can be improved greatly.

Based on the proposed framework, a S/W system was implemented in the PC environment, and it is being used to conduct research on sunscreen products. If the system is made available to consumers, it will help increase the correct usage of sunscreen and reduce the risk of erythema. In the future work, we plan to include the UV index in our system to increase system usability.

CONFLICT OF INTEREST

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Li Z, Kim MA, Kim E, Jung YC, Kim JJ, Shin H‐S. Dynamic visualization of ultraviolet dose on skin with sunscreen applied using minimum erythema dose. Skin Res Technol. 2022;28:614–622. 10.1111/srt.13176

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.