Skin condition evaluation program and skin condition evaluation system

Abstract

To provide a technology that estimates a percutaneous water transpiration quantity accurately from a skin surface image to enable evaluation of a skin condition.SOLUTION: A method for evaluating a skin condition includes: an acquisition step of acquiring a skin surface image of a subject; a calculation step of calculating structure information and color information of a skin surface of the subject from the skin surface image; an estimation step of estimating a percutaneous water transpiration quantity of the subject on the basis of the calculated structure information and color information; and an evaluation step of evaluating a skin condition of the subject on the basis of the estimated percutaneous water transpiration quantity.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a skin condition evaluation method for estimating a subject's transepidermal water loss from an image of the subject's skin surface and evaluating the subject's skin condition based on the estimated transepidermal water loss. [Background technology]

[0002] There is a technique for extracting predetermined parameters from a skin image and presenting parameters related to the skin function of a subject (for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2020 / 189754 Summary of the Invention [Problem to be solved by the invention]

[0004] In the case of Patent Document 1, a feature vector based on topological information of a skin image is extracted, and a parameter related to skin function is presented using an estimation model in which the extracted feature vector is associated with a parameter related to skin function. However, in the course of research, when estimating transepidermal water loss as a parameter related to skin function, there is a risk that the accuracy of the estimated transepidermal water loss is insufficient if only one type of value based on a skin image is extracted and used, and there is room for improvement.

[0005] The present invention has been made in consideration of the above problems, and relates to a technique for extracting multiple types of information from a skin image of a subject and estimating the transepidermal water loss using the multiple types of information extracted, thereby enabling accurate estimation of the transepidermal water loss. In this specification, "skin condition evaluation" refers to evaluation of the apparent state of the skin surface for non-medical purposes, and includes evaluation that can be performed by evaluators other than experts. [Means for solving the problem]

[0006] The present invention relates to a skin condition evaluation method including an acquisition step of acquiring a skin surface image of a subject, a calculation step of calculating structural information and color information of the subject's skin surface from the skin surface image, an estimation step of estimating a transepidermal water loss of the subject based on the calculated structural information and color information, and an evaluation step of evaluating the skin condition of the subject based on the estimated transepidermal water loss.

[0007] The present invention also relates to a skin condition evaluation device including an imaging means for imaging a skin surface image of a subject, a calculation means for calculating structural information and color information of the subject's skin surface from the skin surface image, an estimation means for estimating a transepidermal water loss of the subject based on the calculated structural information and color information, and an evaluation means for evaluating the skin condition of the subject based on the estimated transepidermal water loss. Effect of the Invention

[0008] According to the method provided by the present invention, structural information and color information of the skin surface are calculated from an image of the subject's skin surface, the subject's transepidermal water loss is estimated based on the calculated structural information and color information, and the subject's skin condition is evaluated based on the estimated transepidermal water loss, thereby making it possible to estimate the transepidermal water loss with high accuracy and perform an accurate evaluation of the skin condition. [Brief description of the drawings]

[0009] [Figure 1] (a) is a skin surface image of skin with low transepidermal water loss, (b) is a skin surface image of skin with higher transepidermal water loss than (a), and (c) is a skin surface image of skin with high transepidermal water loss. [Diagram 2] 1 is a flowchart of a skin condition evaluation method. [Diagram 3] 13 is a flowchart for calculating structural information of the skin surface. [Figure 4] 13 is a flowchart for calculating color information of the skin surface. [Diagram 5](a) is a graph showing the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (threshold 50) as the explanatory variable. Similarly, (b) is a graph showing the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (threshold 100), (c) is a graph showing the relationship between structural information (threshold 150), (d) is a graph showing the relationship between color information (L*), (e) is a graph showing the relationship between color information (a*), and (f) is a graph showing the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (threshold 50) as the explanatory variable. [Figure 6] Graph (a) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 100) as an explanatory variable. Similarly, graph (b) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 100, 150) and graph (c) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 150) as an explanatory variable. [Figure 7] Graph (a) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (threshold 50) and color information (L*, a*, b*) as explanatory variables. Similarly, graph (b) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (threshold 100) and color information (L*, a*, b*), and graph (c) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (threshold 150) and color information (L*, a*, b*) as explanatory variables. [Figure 8] Graph (a) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 100) and color information (L*, a*, b*) as explanatory variables. Similarly, graph (b) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 100, 150) and color information (L*, a*, b*), and graph (c) shows the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 150) and color information (L*, a*, b*) as explanatory variables. [Figure 9] 1 is a graph showing the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 100), color information (L*, a*, b*), and sensory information as explanatory variables. [Figure 10] FIG. 1 is a conceptual diagram of a skin condition evaluation device. [Figure 11] FIG. 1 is a conceptual diagram of a skin condition evaluation system. [Figure 12] 1 is a graph showing the relationship between actual transepidermal water loss and estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 100, 150), color information (L*, a*, b*), sensory information, and stratum corneum moisture content as explanatory variables. [Figure 13-1] (a) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information of low-quality images (threshold 50) as an explanatory variable, (b) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information of low-quality images (threshold 100) as an explanatory variable, (c) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information of low-quality images (threshold 150) as an explanatory variable, (d) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information obtained only from the saturation of low-quality images (threshold 50) as an explanatory variable, and (e) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information obtained only from the saturation of low-quality images (threshold 100) as an explanatory variable, and (f) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information obtained only from the saturation of low-quality images (threshold 150) as an explanatory variable. [Figure 13-2] Graph (g) shows the relationship between actual and estimated transepidermal water loss using a regression equation with color information (L*) of the low-quality image as an explanatory variable. Similarly, graph (h) shows the relationship between actual and estimated transepidermal water loss using a regression equation with color information (a*) of the low-quality image as an explanatory variable, and graph (i) shows the relationship between actual and estimated transepidermal water loss using a regression equation with color information (b*) of the low-quality image as an explanatory variable. [Figure 14](a) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 100) obtained from low-quality images as explanatory variables, (b) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information (thresholds 100, 150) obtained from low-quality images as explanatory variables, (c) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 150) obtained from low-quality images as explanatory variables, (d) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 100) obtained only from the saturation of low-quality images as explanatory variables, and (e) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information (thresholds 100, 150) obtained only from the saturation of low-quality images as explanatory variables, and (f) is a graph showing the relationship between the actual transepidermal water loss and the estimated transepidermal water loss using a regression equation with structural information (thresholds 50, 150) obtained only from the saturation of low-quality images as explanatory variables. [Figure 15] Graph (a) shows the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 50) and color information (L*, a*, b*) obtained from low-resolution images as explanatory variables. Similarly, graph (b) shows the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 100) and color information (L*, a*, b*) obtained from low-resolution images as explanatory variables, and graph (c) shows the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 150) and color information (L*, a*, b*) obtained from low-resolution images as explanatory variables. (d) is a graph showing the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 50) and color information (L*, a*, b*) obtained only from the saturation of low-resolution images as explanatory variables. Similarly, (e) is a graph showing the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 100) and color information (L*, a*, b*) obtained only from the saturation of low-resolution images as explanatory variables, and (f) is a graph showing the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 150) and color information (L*, a*, b*) obtained only from the saturation of low-resolution images as explanatory variables. [Figure 16]Graph (a) shows the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 50, 100) and color information (L*, a*, b*) obtained from low-resolution images as explanatory variables. Similarly, graph (b) shows the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 100, 150) and color information (L*, a*, b*) obtained from low-resolution images as explanatory variables. Graph (c) shows the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 50, 150) and color information (L*, a*, b*) obtained from low-resolution images as explanatory variables. (d) is a graph showing the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 50, 100) and color information (L*, a*, b*) obtained only from the saturation of low-resolution images as explanatory variables. Similarly, (e) is a graph showing the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 100, 150) and color information (L*, a*, b*) obtained only from the saturation of low-resolution images as explanatory variables. (f) is a graph showing the relationship between actual and estimated transepidermal water loss using a regression equation with structural information (threshold 50, 150) and color information (L*, a*, b*) obtained only from the saturation of low-resolution images as explanatory variables. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. First, the transepidermal water loss estimated by the skin condition evaluation method of this embodiment (hereinafter, sometimes referred to as this method) will be described with reference to FIGS. 1(a), (b), and (c). The skin plays an important role (barrier function) as a boundary between the inside of the body and the outside, and if the barrier function of the stratum corneum is impaired, skin problems such as dryness and rough skin occur. Measuring the amount of transepidermal water loss is widely used as a method for evaluating the barrier function of the skin. Transepidermal water loss (TEWL) is the amount of water that evaporates from the body through the stratum corneum. It is different from sweat secreted from sweat glands due to an increase in body temperature, exercise, mental stimulation, etc. For example, when the skin becomes rough, an increase in transepidermal water loss is observed, making it a useful indicator for judging the condition of the skin. Transepidermal water loss can be measured, for example, by using a measuring device in which two highly sensitive temperature and humidity sensors are placed at a fixed distance inside a cylindrical chamber at the tip of a probe, and assuming that water evaporating from the skin surface diffuses according to Fick's law, the temperature and humidity differences in the water passing through each sensor are measured, and the amount of water loss is calculated from these values.

[0011] 1(a)-(c) are skin images of five subjects with different transepidermal water evaporation rates. FIG. 1(a) is a skin image of a subject with low transepidermal water evaporation rate, FIG. 1(b) is a skin image of a subject with a transepidermal water evaporation rate higher than FIG. 1(a) and lower than FIG. 1(c), and FIG. 1(c) is a skin image of a subject with high transepidermal water evaporation rate. As is clear from FIG. 1(a)-(c), the skin image of the subject with high transepidermal water evaporation rate has more unevenness on the skin surface than the skin image of the subject with low transepidermal water evaporation rate. In addition, although it is difficult to grasp from FIG. 1(a)-(c), the skin image of the subject with high transepidermal water evaporation rate has more unevenness and reddish skin than the skin image of the subject with low transepidermal water evaporation rate. When the transepidermal water evaporation rate is high, it is suggested that the barrier function of the stratum corneum is reduced, resulting in a rough skin state. Similarly, in diseases with reduced barrier function, the transepidermal water evaporation rate is high, causing inflammation on the skin surface, and the skin surface in a rough or inflamed state is reddish or reddish-black in color, and furthermore, shadows are generated due to unevenness. Thus, it can be seen that the shape and color of the skin surface differ depending on the amount of transepidermal water evaporation. In this method, the amount of transepidermal water evaporation is estimated from a captured skin surface image using points where the shape and color of the skin surface differ depending on the amount of transepidermal water evaporation. Therefore, since the user can easily estimate the amount of transepidermal water evaporation compared to measuring with a measuring device, the amount of transepidermal water evaporation can be continuously estimated in daily life, and changes in the amount of transepidermal water evaporation can also be confirmed over time.

[0012] The method for evaluating a skin condition according to the present embodiment (the present method) will be described with reference to a flowchart shown in FIG. As shown in FIG. 2, this method includes an acquisition step (step S100), a calculation step (steps S110 and S120), an estimation step (step S130), and an evaluation step (step S140).

[0013] The step (step S100) is an acquisition step of acquiring a skin surface image of a subject. The skin surface may be any part of the subject's body, but is preferably the face, neck, arms, and back of the hand, which are easily affected by the amount of transepidermal water evaporation. The skin surface image is an image captured by a camera, or an image captured using a general camera application software provided in a mobile phone or tablet terminal, or a photography application software capable of acquiring RAW data as it is. In this embodiment, image data captured in a close-up state using a photography application software capable of acquiring RAW data as it is is used, but is not limited to this. In addition, the image may be an image captured and acquired when estimating the amount of transepidermal water evaporation by this method, or an image captured in the past may be acquired. By estimating using images captured at different times, it is possible to evaluate changes in skin condition over time. The method of acquiring the skin surface image may be any method, such as directly acquiring the captured skin surface image data (equipped with a photography function in the device that executes this method), acquiring via a network, or acquiring by transferring using a predetermined medium.

[0014] The step (step S110) is a step of calculating structural information of the subject's skin surface from the acquired skin surface image. This step will be described with reference to FIG. In the step (step S111), the acquired skin surface image is trimmed to a predetermined size. In the first embodiment, the image is trimmed to a size of 1200×1200 [pix], and in the second embodiment, the image is trimmed to a size of 100×100 [pix]. In the first embodiment, the image is trimmed to a size of 1200×1200 [pix], and in the second embodiment, the image is trimmed to a size of 100×100 [pix], but the size is not limited to this. Depending on the acquired skin surface image, trimming may not be necessary. The acquired skin surface image is required to have an area of ​​at least 6 mm×6 mm in order to measure the unevenness of the texture from the skin surface image and to perform subsequent processing, and is preferably smaller than 60 mm×45 mm. In the step (step S112), the trimmed image is subjected to a smoothing process and the like to extract a feature region. In this embodiment, as structural information of the skin surface, an uneven region formed by smoothing the moisture content of the skin is extracted. The uneven region formed by smoothing the skin moisture content is not a global unevenness such as the nose, philtrum, or lips, but a local uneven region formed on the skin surface, such as scales that occur when the epidermis' proliferation ability decreases due to a decrease in the moisture content of the skin and the keratin thickens defensively, texture that is a mesh state of the skin's surface skin grooves and ridges, desquamation that is a phenomenon in which the stratum corneum that forms the outermost layer of the epidermis thickens and peels off from the epidermis in various shapes, and infiltration and thickening that occurs when skin cells are produced excessively and the skin piles up thickly. In this embodiment, the scales and texture regions that change due to an increase in transepidermal water loss are calculated as structural information of the skin surface. A low-pass filter (smoothing filter) such as a Gaussian filter is applied to the trimmed image, and subtraction processing is performed between the obtained low-frequency image and the trimmed image to obtain a difference image. The obtained difference image is binarized with a predetermined threshold value, so that an image showing scales and texture can be extracted. In the binarized image, white pixels are parts showing scales and texture. The predetermined threshold value will be explained. In this embodiment, two patterns of predetermined threshold values ​​are set. Although the details will be described later, this is because the amount of transepidermal water loss can be calculated with higher accuracy by using structural information of the skin surface calculated from each of the two patterns of threshold values. In this embodiment, the brightness level of each pixel in the difference image is expressed in 8 bits (0 to 255), with 0 being black and 255 being white. Then, among 0 to 255, the first predetermined threshold value is set to 50, and 50 or less is determined as black, and 51 or more is determined as white (binarized), that is, 51 or more is determined as the area of ​​scales and texture. The second predetermined threshold is set to 100, and values ​​below 100 are judged as black, and values ​​above 101 are judged as white, i.e., values ​​above 101 are scale and texture areas. It is also possible to calculate the transepidermal water loss with high accuracy by performing intermediate processing on the trimmed image before performing smoothing processing or the like in the step (step S112). In the second embodiment, the trimmed image was subjected to smoothing processing after extracting only the chroma (C) element in the Munsell color system. In the step (step S113), the areas of the scale and texture regions are calculated based on the determination results in step S112. Note that in this embodiment, since two patterns of predetermined thresholds are set, the areas of the scale and texture regions at the first threshold and the areas of the scale and texture regions at the second threshold are calculated.

[0015] The step (step S120) is a step of calculating color information of the subject's skin surface from the acquired skin surface image. This step will be described with reference to FIG. In the step (step S121), the acquired skin surface image is trimmed to a predetermined size. In this embodiment, the image is trimmed to a size of 1200 x 1200 [pix]. Since this step is the same as step S111, the image trimmed in step S111 may be used, and this step does not need to be performed again. In the step (step S122), the trimmed image is subjected to a L which indicates brightness in a color space. * , chromaticity indicating hue and saturation is a * , b * The average value of each of the three is calculated. The color information of each pixel of the cropped image data is calculated as L * a * b * By converting it into a color system, * Value, a * Value, b * It is possible to obtain numerical values ​​for each of the values. This conversion is carried out using known image processing software.

[0016] The step (step 130) is a step of estimating the transepidermal water loss of the subject based on the calculated structural information and color information. In the first and second embodiments, the estimation is performed using a multiple regression equation with the structural information and color information as explanatory variables and the transepidermal water loss as a target variable. The multiple regression equation may be calculated and acquired in advance based on the actual transepidermal water loss, structural information, and color information acquired from a plurality of subjects. The relational information acquisition step of acquiring the multiple regression equation may be performed in step S130 or may be performed before step S130. The structural information and color information acquired from a plurality of subjects, which are necessary to acquire the multiple regression equation, are the same as those in the method of steps S100 to S120.

[0017] The step (step S140) is a step of evaluating the skin condition based on the estimated transepidermal water evaporation. By evaluating the transepidermal water evaporation, the dryness state of the skin, i.e., the barrier function, can be evaluated. As an evaluation method, the skin condition of the subject is evaluated in multiple stages by comparing with the transepidermal water evaporation of a healthy person shown in known literature. In this way, it is possible to grasp the skin condition of the subject from the transepidermal water evaporation estimated from the captured image. In addition, as another evaluation method, the transepidermal water evaporation in the initial state (predetermined time) of the subject is used as a reference, and the skin condition (skin barrier condition) is evaluated by the relative change amount such that if the estimated transepidermal water evaporation increases relative to the reference transepidermal water evaporation, the skin condition tends to worsen, and if the estimated transepidermal water evaporation decreases, the skin condition tends to improve. Note that the subject of evaluation may be a computer or a person. In this way, it is possible to grasp the change in the skin condition (skin barrier condition) of the subject by evaluating the transepidermal water evaporation estimated from the captured image over time.

[0018] Here, the multiple regression equation used to calculate the transepidermal water loss will be described in detail. In the first and second embodiments, the actual transepidermal water loss, structure information of the skin surface, and color information of 49 or 85 subjects were obtained to calculate the multiple regression equation. Three types of predetermined thresholds, 50, 100, and 150, were used in step S112. Then, the optimal explanatory variables were determined using a stepwise method to calculate the optimal multiple regression equation. As described above, the first embodiment used an image trimmed to a size of 1200 x 1200 [pix], and the second embodiment used an image trimmed to a size of 100 x 100 [pix].

[0019] 1) The explanatory variables are structural information calculated from one type of threshold or one type of color information. <Embodiment 1> 5(a) to (f) show graphs in which the estimated transepidermal water loss estimated by a regression equation in the first embodiment, calculated from one type of threshold value or one type of color information, is used as an explanatory variable, is plotted on the vertical axis, and the actually measured transepidermal water loss of the subject is plotted on the horizontal axis. In the graphs, the coefficient of determination (R 2 ) is shown. As is clear from Figure 5(a)-(f), in the case of one explanatory variable, the value of the coefficient of determination is the best, and the color information (L * ) is 0.39, and color information (a * ) is 0.38, structural information (threshold 50) is 0.33, structural information (threshold 100) is 0.24, structural information (threshold 150) is 0.16, color information (a * ) is 0.05, so it is not possible to calculate a precise regression equation for any of the explanatory variables. <Embodiment 2> 13-1(a) to (c) show graphs in which the estimated transepidermal water loss estimated by the regression equation when structural information calculated from one type of threshold in the second embodiment is used as an explanatory variable, is plotted on the vertical axis, and the actual transepidermal water loss of the subject is plotted on the horizontal axis. FIG. 13-1(d) to (f) show graphs in which the estimated transepidermal water loss estimated by the regression equation when structural information calculated from one type of threshold in a similar manner is used as an explanatory variable, with only the saturation of the image extracted, is plotted on the vertical axis, and the actual transepidermal water loss of the subject is plotted on the horizontal axis. FIG. 13-2(g) to (i) show graphs in which the estimated transepidermal water loss estimated by the regression equation when one type of color information is used as an explanatory variable, is plotted on the vertical axis, and the actual transepidermal water loss of the subject is plotted on the horizontal axis. In the figures, the coefficient of determination (R 2 ) is shown. When calculating the structural information and color information in the same manner as in the first embodiment, depending on the image quality, as shown in Figs. 13-1(a) to (c) and Figs. 13-2(g) to (i), in the case of one explanatory variable, the value of the coefficient of determination is the best, and the color information (a *), the coefficient of determination was 0.35 when using the structural information (threshold 50) and 0.02, the structural information (threshold 100) and 0.08, and the structural information (threshold 150) were 0.00, so a highly accurate regression equation could not be calculated for any of the explanatory variables. In particular, the coefficient of determination for the structural information was very low, and it was thought that it was not able to properly pick up local uneven areas. In such cases, if only the chroma (C) element in the Munsell color system is extracted, scale and texture information can be obtained stably, and it was found that this improves to 0.09 for the structural information (threshold 50), 0.29 for the structural information (threshold 100), and 0.12 for the structural information (threshold 150), as shown in Figures 13-1(d) to (f).

[0020] 2) The explanatory variables are structural information calculated from any two types of thresholds. <Embodiment 1> 6(a) to (c) are graphs showing the estimated transepidermal water loss estimated by multiple regression equations in the first embodiment when the structural information calculated from each of two thresholds is used as an explanatory variable, and the measured transepidermal water loss of the subject is shown on the horizontal axis. FIG. 6(a) shows the case where the structural information when the thresholds are 50 and 100 is used as an explanatory variable, FIG. 6(b) shows the case where the structural information when the thresholds are 100 and 150 is used as an explanatory variable, and FIG. 6(c) shows the case where the structural information when the thresholds are 50 and 150 is used as an explanatory variable. The coefficient of determination (R 2 ) is shown. As is clear from Figures 6(a) to (c), when two types of structural information are used as explanatory variables, the best values ​​of the coefficient of determination are 0.35 for structural information (thresholds 50 and 100) and structural information (thresholds 50 and 150), and 0.29 for structural information (thresholds 100 and 150). Therefore, although the coefficient of determination is higher than when one explanatory variable is used, a precise multiple regression equation could not be calculated for any of the explanatory variables. <Embodiment 2> Fig. 14(a) to (c) show graphs with the estimated transdermal water loss amount estimated by the regression equation when the structure information calculated from each of the two types of thresholds is used as an explanatory variable on the vertical axis and the actually measured transdermal water loss amount of the subject on the horizontal axis in Embodiment 2. Fig. 14(d) to (f) show graphs with the estimated transdermal water loss amount estimated by the regression equation when the structure information calculated from two types of thresholds in the same manner is used as an explanatory variable with only the saturation of the image extracted on the vertical axis and the actually measured transdermal water loss amount of the subject on the horizontal axis. In the figures, the coefficient of determination (R 2 ) is shown. As is clear from Fig. 14(a) to (c), in the case of two explanatory variables, the value of the coefficient of determination is the best, which is 0.11 for the structure information (thresholds 100 and 150), 0.08 for the structure information (thresholds 50 and 100), and 0.07 for the structure information (thresholds 50 and 150). Therefore, the value of the coefficient of determination has increased compared to the case of one explanatory variable, but a regression equation with good accuracy has not been calculated for any of the explanatory variables. Even when only the element of saturation (C) in the Munsell color system is extracted, as shown in Fig. 14(d) to (f), the value of the coefficient of determination is the best, which is 0.32 for the structure information (thresholds 100 and 150), 0.30 for the structure information (thresholds 50 and 100), and the improvement only reaches 0.18 for the structure information (thresholds 50 and 150).

[0021] 3) Set the explanatory variables as the structure information calculated from any one type of threshold and the four color information (L * , a * , b * ). <Embodiment 1> Fig. 7(a) to (c) show graphs with the estimated transdermal water loss amount estimated by the multiple regression equation when the structure information calculated from one type of threshold and the four color information (L * , a * , b * ) are used as explanatory variables in Embodiment 1 on the vertical axis and the actually measured transdermal water loss amount of the subject on the horizontal axis. In the figures, the coefficient of determination (R 2 ) is shown. As is clear from Fig. 7(a) to (c), for one type of structure information and the four color information (L * , a * , b *) are used as explanatory variables, the best coefficient of determination value is 0.53 for structure information (threshold 50) and color information, 0.50 for structure information (threshold 100) and color information, and 0.49 for structure information (threshold 100) and color information. Therefore, the coefficient of determination value is higher than when the explanatory variable is only structure information. In particular, in the case of the four explanatory variables of structure information (threshold 50) and color information, a precise multiple regression equation can be calculated. <Embodiment 2> 15(a) to (c) show structure information and color information (L * , a * , b * 15(d) to (f) show graphs in which the estimated transepidermal water loss estimated by a multiple regression equation when the four explanatory variables are the above four items, 1) the amount of transepidermal water loss of the subject, and 2) the amount of transepidermal water loss of the subject, are plotted on the vertical axis. * , a * , b * ) are used as explanatory variables, the vertical axis shows the estimated transepidermal water loss estimated by the multiple regression equation, and the horizontal axis shows the actual transepidermal water loss of the subject. 2 ) is shown. As is clear from Fig. 15(a) to (c), one type of structural information and color information (L * , a * , b * ) are used as explanatory variables, the best coefficient of determination is 0.44 for structural information (threshold 100) and color information, 0.43 for structural information (threshold 50) and color information, and 0.42 for structural information (threshold 150) and color information. Therefore, the coefficient of determination is higher than when the explanatory variable is only structural information, and in particular, in the case of four explanatory variables of structural information (threshold 100) and color information, a highly accurate multiple regression equation can be calculated. Furthermore, as shown in Figures 15(d) to (f), when only saturation is extracted, the best coefficient of determination is 0.49 for structural information (threshold 100) and color information, 0.43 for structural information (threshold 50) and color information, and 0.48 for structural information (threshold 150) and color information, indicating that an even more accurate multiple regression equation can be calculated.

[0022] 4) The structural information and color information (L * , a * , b * ) <Embodiment 1> 8(a) to (c) show structure information and color information (L * , a * , b * ) are used as explanatory variables, the vertical axis shows the estimated transepidermal water loss estimated by multiple regression equation, and the horizontal axis shows the actual transepidermal water loss of the subject. 2 ) is shown. As is clear from Fig. 8(a) to (c), two types of structural information and color information (L * , a * , b * ) are used as explanatory variables, the best values ​​of the coefficient of determination are 0.58 for structure information (threshold 50, 100) and color information, 0.55 for structure information (threshold 100, 150) and color information, and 0.54 for structure information (threshold 50, 150) and color information. Therefore, when the explanatory variables are one type of structure information and one type of color information (L * , a * , b * ) The coefficient of determination is higher than when the threshold value was 50 and 100. In particular, when the five explanatory variables for structural information (threshold values ​​of 50 and 100) and color information were used, a multiple regression equation with good accuracy was calculated. <Embodiment 2> 16(a) to (c) show structure information and color information (L * , a * , b * 16(d) to (f) show graphs in which the estimated transepidermal water loss estimated by a multiple regression equation when the five explanatory variables are the above-mentioned three-dimensional image structure, the color information (L, * , a * , b *) are used as explanatory variables, the vertical axis shows the estimated transepidermal water loss estimated by the multiple regression equation, and the horizontal axis shows the actual transepidermal water loss of the subject. 2 ) is shown. As is clear from Fig. 16(a) to (c), two types of structural information and color information (L * , a * , b * ) are used as explanatory variables, the coefficient of determination is 0.44 for structure information (threshold 50, 100) and color information, structure information (threshold 100, 150) and color information, and structure information (threshold 50, 150) and color information. Therefore, the explanatory variables are one type of structure information and one type of color information (L * , a * , b * 16(d) to (f), when only saturation is extracted, the coefficient of determination is the best, at 0.51 for structural information (threshold 100, 150) and color information, 0.49 for structural information (threshold 50, 100) and color information, and 0.48 for structural information (threshold 50, 150) and color information, indicating that a multiple regression equation with even higher accuracy can be calculated.

[0023] These results suggest that the two types of structural information and color information (L * , a * , b * It was found that the transepidermal water loss rate can be estimated with high accuracy by using a multiple regression equation with the five explanatory variables. In particular, it was found that the thresholds for the structural information were 50 and 100, which had the best coefficient of determination. In either case, the coefficient of determination by the regression equation including the structural information using the threshold value of 50 tends to be high. In this embodiment, the brightness of one pixel is expressed in 0 to 255 stages (black to white), and it was found that a more accurate regression equation can be calculated by using the structural information calculated using the threshold value set closer to black than the intermediate brightness value as an explanatory variable. Therefore, it is preferable to set the threshold value when calculating the structural information closer to black than the intermediate value (127), and when two types of threshold values ​​are set, it is preferable to set the first threshold value to 40 to 60 (40 / 256 to 60 / 256 when the brightness of one pixel is 0 to 255 stages) and the second threshold value to 82 to 102 (82 / 256 to 102 / 256 when the brightness of one pixel is 0 to 255 stages). In other words, it is preferable to set the first threshold value to 15% to 25% and the second threshold value to 32% to 40% for the brightness stages of one pixel. The ratio of the first threshold value and the second threshold value to the brightness level of one pixel can be applied even when the brightness level of one pixel is different. Alternatively, the ratio of the first threshold value to the second threshold value is preferably 1:1.3 to 1:2.6. In addition, when the brightness level of the first threshold value is 100, the brightness level of the second threshold value is preferably 125 to 300, more preferably 150 to 250. Furthermore, when comparing the results obtained when using the image itself cropped to a size of 100 x 100 [pix] in embodiment 2 with the results obtained when using an image cropped to a size of 100 x 100 [pix] and extracting only the chroma (C) element in the Munsell color system, it was found that the results obtained when using an image cropped to a size of 100 x 100 [pix] and extracting only the chroma (C) element showed improved accuracy in the estimated transepidermal water loss regardless of the explanatory variable used. In addition, when the image itself cropped to a specified size was used, the accuracy of the estimated transepidermal water loss was higher when a high-quality image was used. However, considering that ordinary users take pictures of their own skin in their daily lives and easily understand changes in their skin condition on a daily basis using the skin images, it was found that even when an image cropped to a size of, for example, 100 x 100 [pix] is used, it is possible to estimate transepidermal water loss with sufficient accuracy by performing a process to extract only the saturation (C) element and taking into account the combination of explanatory variables. In this embodiment, two types of thresholds are used, but the present invention is not limited to this, and three or more types of thresholds may be used. Even in this case, it is preferable that the thresholds include a value closer to black than the intermediate value of the brightness level of one pixel.

[0024] Next, a case will be described in which sensory information, which is a numerical value of the sensation felt by the subject on the surface or inside of the skin, is added to the explanatory variables. Sensory information is information that is a numerical value of the level (degree) of sensations felt by the subject on the surface or inside of the skin, such as itching, pain, heat, and tightness. It is preferable that the sensation is caused by an increase in the amount of transepidermal water loss. A questionnaire may be taken about the level at which any of these sensations are felt, or a questionnaire may be taken about the level at which any one of these sensations is felt. In this embodiment, as sensory information, a questionnaire was taken of subjects to determine which of five levels of sensations they felt on the skin surface: no itching at all, almost no itching, neither, a little itching, and a lot of itching, and the questionnaire results were quantified and used as explanatory variables. Sensory information obtained in the same manner from multiple subjects was added to explanatory variables to calculate a multiple regression equation. Based on the above results, structural information (threshold value 50, 100), color information (L * , a * , b * ), and sensory information as six explanatory variables. The estimated transepidermal water loss estimated by multiple regression equation is plotted on the vertical axis against the actual transepidermal water loss of the subjects on the horizontal axis. The coefficient of determination (R 2 ) is shown. As is clear from FIG. 9, the structure information (thresholds 50 and 100) and color information (L * , a * , b * ), and the six explanatory variables of sensory information, the coefficient of determination was 0.65, resulting in a highly accurate multiple regression equation. Note that since the sensory information is highly likely to vary depending on the subject's subjective opinion, it is necessary to ensure that subjects can respond appropriately to the questionnaire result levels. When sensory information is not included (structure information (threshold 50, 100) and color information (L * , a *, b * )), the coefficient of determination is highly accurate, so a multiple regression equation may be used according to the subject's understanding (for example, a multiple regression equation excluding sensory information in the case of a child who is thought to have difficulty answering with a certain sense, and a multiple regression equation including sensory information in the case of an adult who is thought to be able to answer with a certain sense).

[0025] <Skin condition evaluation device> Next, a skin condition evaluating device 10 equipped with the above-mentioned skin condition evaluating method will be described. FIG. 10 shows a conceptual diagram of the skin condition evaluating device 10. The skin condition evaluation device 10 comprises an imaging means 11, a calculation means 12, an estimation means 13, and an evaluation means 14. In this embodiment, the skin condition evaluation device 10 is a mobile terminal equipped with an imaging means 11 (camera) and a central processing unit (not shown), such as a smartphone or a tablet terminal, but may also be a camera equipped with a central processing unit. The image capturing means 11 can be a camera built into a smartphone or a tablet terminal, but needs to be an RGB camera in order to calculate color information of the skin surface. The calculation means 12 is a means for executing the process of calculating the structural information and color information of the skin surface of the subject described above, and is provided in the central processing unit. The estimation means 13 is a means for executing the process of estimating the transepidermal water loss of the subject described above, and is provided in the central processing unit. The evaluation means 14 is a means for executing the process of evaluating the skin condition of the subject based on the transepidermal water loss described above, and is provided in the central processing unit. The skin condition evaluation device 10 is also provided with a storage means 15, which stores a computer program for executing the above-mentioned skin condition evaluation method. The skin condition evaluation device 10 is also provided with a display means 16. The display means 16 is a display device provided in the mobile terminal 10. By displaying the results of the skin condition evaluation of the subject by the evaluation means 14 on the display means 16, it becomes easier for the subject to understand the skin condition. In addition, although not shown, the skin condition evaluation device 10 is equipped with input devices such as a keyboard (including a touch keyboard) and a pointing device, as well as an output device required for outputting to an external device such as a printer.

[0026] <Skin condition evaluation system> Next, a skin condition evaluating system including the above-mentioned skin condition evaluating method will be described. FIG. 11 shows a conceptual diagram of a skin condition evaluating system 100. The skin condition evaluation system 100 includes a mobile terminal 110 and a server 120 . Like the mobile terminal 10, the mobile terminal 110 is a smartphone or tablet terminal equipped with an image capturing means 111 (camera) and a central processing unit (not shown), or a camera equipped with a central processing unit. It is preferable that the mobile terminal 110 is equipped with a display means 116 so that the evaluation results can be easily confirmed. The display means 116 is a display device equipped in the mobile terminal 110. In addition, although not shown, the mobile terminal 110 is equipped with an input / output means for transmitting a skin surface image captured by the image capturing means 111 to a server 120 via the Internet or the like, and receiving an estimated transepidermal water loss and evaluation results from the server 120. The server 120 stores a computer program for executing the above-mentioned skin condition evaluation method and a skin surface image captured by the image capture means 111. The server 120 includes a calculation means 122, an estimation means 123, an evaluation means 124, and a storage means 125. The calculation means 122 is a means for executing the processing content of a calculation step for calculating the structural information and color information of the skin surface of the subject described above. The estimation means 123 is a means for executing the processing content of an estimation step for estimating the transepidermal water loss of the subject described above. The evaluation means 124 is a means for executing the processing content of an evaluation step for evaluating the skin condition of the subject based on the above-mentioned transepidermal water loss. The storage means 125 is a means for storing a computer program for executing the skin condition evaluation method and a skin surface image captured by the image capture means 111. In addition, although not shown, the server 120 includes an input / output means for transmitting the transepidermal water loss estimated by the server 120 and the evaluation result to the mobile terminal 110 via the Internet or the like and receiving the captured skin surface image from the mobile terminal 110. In this way, by having the server 120 execute a computer program for executing the skin condition evaluation method, it is possible to reduce the processing load on the user's mobile terminal 110. Also, by storing the captured skin surface images in the server 120, all of the users' skin surface images can be collectively stored, so that the skin surface images can be used to improve the algorithm in the computer program for executing the skin condition evaluation method, and more accurate estimation and evaluation can be performed. Also, since the user's past skin surface images, estimated transepidermal water loss, and evaluation results are stored, the user can access the server 120 to check detailed changes over time.

[0027] <Modification> In this embodiment, the area of ​​the uneven area formed in the skin surface image that affects the moisture content of the skin is calculated as the structural information of the skin surface, but this is not limited to this. For example, the area ratio of the uneven area formed on the skin surface to the area other than the uneven area may be calculated and used to estimate the transepidermal water loss. This method also provides the same effect as the area calculation method.

[0028] In this embodiment, the irregularities formed on the skin surface are scales and texture, but are not limited thereto. Other skin conditions that occur on the skin surface due to changes in transepidermal water loss (effects of increased transepidermal water loss) may also be used.

[0029] In this embodiment, the color information for each pixel is stored as L * a * b * By converting it into a color system, * Value, a * Value, b * Although the numerical values ​​for each value were calculated, this is not limiting. For example, it is also possible to convert to the Munsell color system and calculate H (hue), V (value), and C (chroma). As long as the value, hue, and chroma can be calculated, the color system is not important.

[0030] The display result displayed on the display means 16 (display means 116) may be only the estimated evaluation result for the current time, or may be displayed together with the evaluation results of past evaluations. In this way, the subject can take a skin surface image himself, estimate the transepidermal water loss, and grasp the obtained evaluation results over time, so that it becomes possible to easily grasp the change in the skin condition on a daily basis.

[0031] In this embodiment, the sensory information is obtained by asking the subjects to indicate which of the five levels of itching they feel: no itching at all, almost no itching, neutral, a little itching, and very itchy, but this is not limited to the above. The questionnaire may be conducted using a visual analog scale (VAS) that represents the degree of itching, and the questionnaire results may be quantified. In this way, the detailed sensations (itching levels) of the subjects can be reflected in the questionnaire results, and the accuracy of estimating the transepidermal water loss can be improved.

[0032] In this embodiment, the transepidermal water loss is calculated using a multiple regression equation, but is not limited thereto. For example, a regression model such as Ridge, Lasso, Elastic Net, Random Forest, Neural Network, Support Vector Regression, etc. may be created by machine learning based on the structural information and color information of the skin surface acquired from multiple subjects and the measured transepidermal water loss, and the transepidermal water loss may be calculated using the created regression model and the calculated structural information and color information of the skin surface. When creating a regression model by machine learning, sensory information acquired from multiple subjects may also be used to create the regression model, and the transepidermal water loss may be calculated using the created regression model, the calculated structural information and color information of the skin surface, and the acquired sensory information. In addition, the structural information and color information of the skin surface may be calculated from the skin surface image of the subject by machine learning using data learned from the skin surface images, structural information and color information of the skin surface acquired from multiple subjects.

[0033] In this embodiment, structural information, color information, and sensory information of the skin surface are used as explanatory variables, but the present invention is not limited to this. For example, in addition to structural information, color information, and sensory information of the skin surface, the stratum corneum moisture content may be added as an explanatory variable. The stratum corneum moisture content is measured using a known device (e.g., a Corneometer). The structural information (thresholds 50, 100, 150), color information (L * , a * , b * ), sensory information, and stratum corneum moisture content are eight explanatory variables. The estimated transepidermal water loss estimated by multiple regression equation is plotted on the vertical axis, and the actual transepidermal water loss of the subjects is plotted on the horizontal axis. The coefficient of determination (R 2 ) is shown. As is clear from FIG. 12, the structure information (threshold values ​​50, 100, and 150) and color information (L * , a * , b * ), sensory information, and stratum corneum moisture content were used as eight explanatory variables, resulting in a coefficient of determination of 0.67, which was a highly accurate multiple regression equation. Thus, by using stratum corneum moisture content as an explanatory variable to estimate transepidermal water loss, it is possible to estimate a more accurate value. [Explanation of symbols]

[0034] 10 Skin condition evaluation device (mobile terminal) 11. Shooting Method 12 Calculation method 13 Estimation means 14 Evaluation methods 15 Memory means 16 Display means 100 Skin condition evaluation system 110 Mobile terminals 111 Filming Method 116 Display means 120 Servers 122 Calculation Method 123 Estimation means 124 Evaluation methods 125 Memory means

Claims

1. A skin condition evaluation program for evaluating a skin condition of a subject, comprising: On the computer, an acquisition process for acquiring a skin surface image of the subject; A calculation process for calculating structural information and color information of the subject's skin surface from the skin surface image; an estimation process for estimating a transepidermal water loss of the subject based on the calculated structure information and color information; and evaluating the skin condition of the subject based on the estimated transepidermal water loss.

2. 2. The skin condition evaluation program according to claim 1, wherein the structural information is information regarding an area of ​​a local uneven region of the skin in the skin surface image.

3. In the calculation process, the computer sets a plurality of thresholds for determining that an uneven area is formed due to a change in the shape of the skin surface in the skin surface image, the structural information includes first structural information determined based on a first threshold and second structural information determined based on a second threshold different from the first threshold, The skin condition evaluation program according to claim 2 , wherein the estimation process estimates the transepidermal water loss based on the first structural information, the second structural information, and the color information.

4. 4. The skin condition evaluation program according to claim 1, wherein the color information is hue and saturation in the skin surface image.

5. and causing the computer to further execute a process of obtaining a regression equation in which the structural information and the color information obtained from a plurality of subjects are explanatory variables and the transepidermal water loss is a response variable.

4. The skin condition evaluation program according to claim 1, wherein the estimation process calculates the transepidermal water loss using the structural information and the color information of the subject and the regression equation.

6. and further causing the computer to execute a process of acquiring sensory information that quantifies the sensation felt by the subject through the skin surface; The regression equation further includes the sensory information acquired from the plurality of subjects as an explanatory variable, The skin condition evaluation program according to claim 5 , wherein the estimation process calculates the transepidermal water loss using the structural information, the color information, and the sensory information of the subject and the regression equation.

7. The computer includes a storage means, The skin condition evaluation program according to claim 1 , wherein the skin condition evaluation program is stored in the storage means.

8. The skin condition evaluation program described in Claim 1, characterized in that the skin condition evaluation program is stored in a server connected to the Internet.

9. A mobile terminal connected to the Internet; A server connected to the Internet, The mobile terminal, An imaging means for imaging the skin surface of the subject to obtain a skin surface image; an input / output means for transmitting and receiving the skin surface image to and from the server via the Internet; The server, a calculation means for calculating structural information and color information of the subject's skin surface from the skin surface image; an estimation means for estimating a transepidermal water loss of the subject based on the calculated structure information and color information; and an evaluation means for evaluating the skin condition of the subject based on the estimated transepidermal water loss.

10. The skin condition evaluation system described in Claim 9, characterized in that the structural information is information regarding the area of ​​local uneven areas of the skin in the skin surface image.

11. The server provides a plurality of thresholds for determining an uneven area formed due to a change in the shape of the skin surface in the skin surface image, the structural information includes first structural information determined based on a first threshold and second structural information determined based on a second threshold different from the first threshold, The skin condition evaluation system according to claim 10 , wherein the estimation means estimates the transepidermal water loss based on the first structural information, the second structural information, and the color information.

12. A skin condition evaluation system described in any one of claims 9 to 11, characterized in that the color information is hue and saturation in the skin surface image.

13. The server obtains a regression equation in which the structural information and the color information for a plurality of subjects are explanatory variables and the transepidermal water loss is a target variable, 12. The skin condition evaluation system according to claim 9, wherein the estimation means calculates the transepidermal water loss using the structural information and the color information of the subject and the regression equation.

14. The server acquires sensory information that quantifies the sensation felt by the subject from the skin surface, The regression equation further includes the sensory information acquired from the plurality of subjects as an explanatory variable, The skin condition evaluation system according to claim 13 , wherein the estimation means calculates the transepidermal water loss using the structural information, the color information, and the sensory information of the subject and the regression equation.

Images