Elsevier

Measurement

Volume 182, September 2021, 109684
Measurement

Simple detection of absorption change in skin tissue using simulated spectral reflectance database

https://doi.org/10.1016/j.measurement.2021.109684Get rights and content

Highlights

  • Absorption-oriented spectral database is proposed for evaluating skin absorption change.

  • Reflectance spectra in database are simulated using nine-layered skin tissue model.

  • Spectral matching with measured spectra is performed by root mean square error.

  • In-vivo validation experiments were made to compare with multiple regression analysis.

Abstract

We present a method for detection of absorption change in skin tissue on the basis of simulated spectra extracted from an absorption-oriented database. We constructed the database with a moderate number of reflectance spectra which are simulated by the Monte Carlo method using the nine-layered skin model. In the construction of database, change of only absorption coefficients in the nine layers is made with two magnification factors for two groups of layers. For a measured spectrum with an absorption change, a simulated spectrum is being extracted out from the database by the criterion of root mean square error. The absorption coefficients associated with the extracted spectrum are used to specify the absorption change in the measured spectrum. We performed experiments of suntan, pressure-cuff occlusion and hot water immersion and compared the results with those of the multiple regression analysis to show the usefulness of the method for evaluating the change of absorption properties from measured spectra.

Introduction

Importance on optical investigation of human skin has grown by leaps and bounds over the years [1], [2], [3], [4], [5], [6]. Through the analysis of spectral reflectance, knowledge of various skin parameters can be obtained regarding skin conditions such as color, diseases and aging. It has been generally recognized that alteration in the spectral reflectance is mostly influenced by absorption and scattering of light in skin as they can be approximately associated to the presence of chromophores such as melanin and hemoglobin, and the tissue morphology. As various skin diseases and conditions govern modification in the chromophores and tissue structure, it results in characterizing the spectral reflectance [7], [8], [9], [10]. Thus, the analysis of skin spectral reflectance became one of the principal research elements in dermatology and cosmetology. Most recently, the skin spectroscopic technique is extended to application for monitoring of physiological condition and health status [11], [12]. This kind of application is considered to be advantageous for the remote or on-line inspection systems, especially in the present serious situations of infectious diseases. To realize such systems, the spectral reflectance analysis should be made in the simple, easy to use, and costless manner. However, in general it is quite difficult to find simply and directly relation between skin tissue conditions and measured spectral reflectance, although various methods have been explored to approach this subject [5], [6], [7], [9], [13], [14], [15].

Among the methods, multi-variate analysis of reflectance spectra is probably the most typical approach which has been studied by many groups of researchers in various ways including works in the articles [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [13], [14]. Imai et al. [16] proposed the use of principal component analysis for the assessment of melanin and hemoglobin concentration through image processing. For the same purpose, Tsumura et al. [17] introduced the method of independent component analysis in color images of human face. Shimada et al. [18], [19], [20] presented the applicability of multiple regression analysis (MRA) to evaluating the concentration of melanin and blood in human skin on the basis of modified Beer-Lambert law [21], [22], [23]. This method suffers from non-linearity between chromophore concentrations and corresponding regression coefficients, which was proficiently handled by Nishidate et al. [24], [25] through use of conversion vectors. Konishi et al. [26] and Spigulis et al. [27] also reported the use of MRA for chromophore distribution imaging. These approaches are likely to provide useful results if their regression models are appropriately designed for analysis. However, the models and regression coefficients should be re-constructed in case of different kinds and amounts of changes in the skin conditions. From this point of view, they seem disadvantageous for system extensibility to general purposes including on-line skin monitoring.

Some other methods include solution of the radiative transfer equation with auxiliary function method [28], [29] which has some limitation in application for various skin conditions, and Kubelka-Munk theory [30] which is limited to homogeneous skin tissue model. Zhang et al. [31], [32] presented the idea of using library based spectral fitting for determination of optical properties through the diffusion model. They presented reasonable estimation of nine skin parameters by genetic algorithms, but their skin model consisted of epidermis and dermis layers only. These models generally hold in limited ranges of absorption and scattering changes, owing to their approximation models employed. Thus, they are considered somehow complicated for finding simply change in skin conditions from measured spectra.

As the absorption coefficient corresponds to a product of molar extinction coefficient and molar concentration of chromophore, and the scattering coefficient corresponds to morphology of tissue and cells, these coefficients are considered as key factors that regulate a reflectance spectrum under a certain skin condition. As a flexible approach to obtain various spectral reflectance curves with given values of absorption and scattering coefficients, Monte Carlo method for layered tissue model [33] is well known and widely employed. Meglinski and Matcher [34], [35] reported simulated reflectance spectra using a seven-layered tissue model which provided usefulness for investigating absorption and scattering coefficients in each of layers, but they assumed wavelength independency for scattering coefficients. On the basis of this report, Maeda et al. [36] developed the nine-layered skin tissue model with wavelength-dependent skin parameters (absorption coefficient, scattering coefficient and anisotropy scattering parameter) together with refractive index and thickness of skin layers. This model generated reasonably similar simulated spectra in accordance with measured spectra. By using this approach, we then presented the spectral fitting method for estimation of absorption and scattering coefficients [37]. In this model, we found that a simulated spectrum becomes similar to the measured one through the adjustment of five parameters. However, this adjustment requires the manual iteration process for finding reasonable values for the five parameters. This is quite time consuming and unstable process for obtaining a simulated spectrum very close to the measured spectrum.

In our previous work [38], we proposed the use of a database consisting of various simulated spectra for estimating the absorption and scattering coefficients of a measured spectrum, in which a simulated spectrum that closely resembles with the measured spectrum is extracted out from the spectral database. The absorption and scattering coefficients associated with the extracted spectrum are regarded as estimated absorption and scattering coefficients of the measured spectrum. In the study, we introduced the moderate grouping method [39] for changing values of absorption and scattering coefficients in the nine layers. By this method, we constructed the spectral reflectance database in a reasonable size while utilizing the merit of having nine layers in skin tissue model, and demonstrated possibilities for estimation of absorption and scattering coefficients by simulation study.

The relation between the absorption coefficient and chromophore concentration is clinically proven, and it is regarded as predominantly consequential in both skin care and diagnostics. The relation between the scattering coefficient and tissue morphology (for example, tissue structure, cell structure, and fiber composition such as collagen and elastin) is yet to be clearly identified for clinical usage. For this reason, many researchers have made much effort to mainly focus on evaluating change in the concentration of chromophores such as melanin and hemoglobin (corresponding to absorption change) with their respective techniques [16], [17], [18], [19], [20], [24], [25], [26], [27], which provide useful information on various skin conditions including cancers, pigmentation and tissue metabolism. So, we consider that the database method has much potential to become an effective means for analysis of chromophores concentration, even if the database is used for estimation of only the absorption coefficients in skin tissue.

In this paper, we propose the use of absorption-oriented database which consists of simulated spectra having variation only in absorption coefficients, while restraining the other skin parameters from changing. Although the reflectance spectrum is influenced by various skin parameters like absorption and scattering coefficients and anisotropy factor, refractive index, and thickness, we employ a way to grasp dominantly the absorption change from the spectrum in the constraint of no change on the other parameters. This choice is expected to be effective for simple detection of absorption change in skin tissue. In order to investigate experimentally the feasibility of the proposed method, we performed sun exposure (suntan) experiment for detecting change in melanin concentration or the absorption coefficient in epidermal layer, and pressure-cuff occlusion and hot water immersion experiments for detecting change in hemoglobin concentration or the absorption coefficient in dermal and subcutaneous layers. To discuss the usefulness of the proposed method, we compared the tendency of estimated parameters with those obtained by the MRA method.

Section snippets

Nine-layered skin tissue model and Monte Carlo simulation

Human skin is known to have a complex structure as can be seen in Fig. 1(a). Generally, a skin tissue model is expressed by three layers which are epidermis, dermis and subcutaneous layers. However, this model comes in short in mimicking various skin conditions aptly. One of such examples is that melanin in epidermis and hemoglobin in dermis have inhomogeneous distribution along the vertical direction in skin tissue. We first noticed the seven-layered model which was developed in the preceding

Method of multiple regression analysis

To assert the effectiveness for analyzing absorption change in skin tissue by the database (DB) method, we also performed multiple regression analysis for the measured spectra obtained by experiments conducted in Section 4. Here, we followed the method of Nishidate et al. [24]. According to the modified Lambert-Beer law [20], [21], [22], an absorbance spectrum A(λ) = -log10 R(λ) can be expressed as,Aλ=Cmleλεmλ+Cohldλεohλ+Cdhldλεdhλ+Sλ,where C and ε(λ) are the molar concentration and molar

Experimental setup

To measure a reflectance spectrum, we used the experimental system shown in Fig. 3 [35]. Light from a halogen lamp (Hayashi-Repic, Tokyo, LA-150UX), which covers a visible wavelength range, was focused through a light guide and lens in such a manner that it produces a spot of 4 mm on the surface of a measured sample. Diameter and focal length of lens were 54 and 100 mm, respectively. The diffuse light reflected from the sample, which was placed at a sample port of an integrating sphere

Influence of g, n, and t

In this study, we fixed the values of g, n, and t to be constant for the construction of the database. However, variation in the three parameters have effects on the skin reflectance spectra, because g and n are related to the scattering or reduced scattering properties and t is also to both absorption and scattering properties in terms of the light propagation length. The values of g, n, and t used in this study almost follow the published results of many researchers [40], [51], [55], [56].

Conclusion

We have described the simple construction of absorption-oriented database, in which only absorption coefficients are varied with keeping fixed values in the other parameters, and intended for better detection of the absorption change in skin tissue. Then, we presented the usefulness of this database by performing the three experiments of the suntan, pressure-cuff occlusion and hot water immersion on the forearm of human subjects. In all of these experiments primarily the absorption change due

Disclosures

The authors have no relevant financial interests in this article and no other potential conflicts of interest to disclose.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by JSPS KAKENHI Grant Number 20K04513.

References (66)

  • E. Salomatina et al.

    Optical properties of normal and cancerous human skin in the visible and near-infrared spectral range

    J. Biomed. Opt.

    (2006)
  • G.N. Stamatas et al.

    In vivo documentation of cutaneous inflammation using spectral imaging

    J. Biomed. Opt.

    (2007)
  • G. Zonios et al.

    Melanin absorption spectroscopy: new method for noninvasive skin investigation and melanoma detection

    J. Biomed. Opt.

    (2008)
  • M. Hirose et al.

    Principal component analysis for surface reflection components and structure in facial images and synthesis of facial images for various ages

    Opt. Rev.

    (2017)
  • I. Nishidate et al.

    Simple and affordable imaging of multiple physiological parameters with RGB camera-based diffuse reflectance spectroscopy

    Biomed. Opt. Exp.

    (2020)
  • D. McDuff et al.

    Non-contact imaging of peripheral hemodynamics during cognitive and psychological stressors

    Sci. Rep.

    (2020)
  • D. Fukutomi et al.

    Determination of the scattering coefficient of biological tissue considering the wavelength and absorption dependence of the anisotropy factor

    Opt. Rev.

    (2016)
  • N. Tsumura et al.

    Mapping pigmentation in human skin from multi-channel visible spectrum image by inverse optical scattering technique

    J. Imaging Sci. Technol.

    (2001)
  • F.H. Imai et al.

    Principal component analysis of skin color and its application to colorimetric color reproduction on CRT display and hard copy

    J. Imaging Sci. Technol.

    (1996)
  • N. Tsumura et al.

    Independent-component analysis of skin color image

    J. Opt. Soc. Am. A:

    (1999)
  • M. Shimada et al.

    Explanation of human skin color by multiple linear regression analysis based on the modified Lambert-Beer law

    Opt. Rev.

    (2000)
  • M. Shimada et al.

    Melanin and blood concentration in human skin studied by multiple regression analysis: experiments

    Phys. Med. Biol.

    (2001)
  • M. Shimada et al.

    Melanin and blood concentration in a human skin model studied by multiple regression analysis: assessment by Monte Carlo simulation

    Phys. Med. Biol.

    (2001)
  • Y. Nomura et al.

    Relationship between time-resolved and non-time- resolved Beer-Lambert law in turbid media

    Phys. Med. Biol.

    (1997)
  • Y. Nomura et al.

    Exponential attenuation of light along nonlinear path through the biological model

    Adv. Exp. Med. Biol.

    (1989)
  • M. Hiraoka et al.

    A Monte Carlo investigation of optical path length in inhomogeneous tissue and its application to near-infrared spectroscopy

    Phys. Med. Biol.

    (1993)
  • I. Nishidate et al.

    Estimation of melanin and hemoglobin in skin tissue using multiple regression analysis aided by Monte Carlo simulation

    J. Biomed. Opt.

    (2004)
  • I. Nishidate et al.

    Non-invasive spectral imaging of skin chromophores based on multiple regression analysis aided by Monte Carlo simulation

    Opt. Lett.

    (2011)
  • I. Konishi et al.

    A new optical image for hemoglobin distribution in human skin

    Opt. Rev.

    (2003)
  • J. Spigulis et al.

    Smartphone snapshot mapping of skin chromophores under triple-wavelength laser illumination

    J. Biomed. Opt.

    (2017)
  • C. Magnain et al.

    Skin color modelling using the radiative transfer equation solved by the auxiliary function method

    J. Opt. Soc. Am. A:

    (2007)
  • C. Magnain et al.

    Skin color modelling using the radiative transfer equation solved by the auxiliary function method: inverse problem

    J. Opt. Soc. Am. A:

    (2008)
  • R. Ohtsuki et al.

    Multiple-reflection model of human skin and estimation of pigment concentrations

    Opt. Rev.

    (2012)
  • Cited by (4)

    View full text