The assessment of moderate acne vulgaris face skin using blood perfusion and hyperspectral imaging–A pilot study

Microcirculation is the flow of blood through the smallest vessels in the circulatory system. Capillaries respond to various pathologies much earlier than arteries and veins, the changes which indicate that the disease is already developing. Careful monitoring of the functioning of the capillary system often allows the detection of in vivo disorders at an early stage.


| INTRODUC TI ON
Microcirculation is the flow of blood through the smallest vessels in the circulatory system.It occurs in all tissues and organs except the cornea and is essential for the proper functioning of tissues.One of the most common causes of changes in the microcirculation of the skin is its inflammation. 1,2Capillaries, which are the smallest and most numerous blood vessels in the human body, respond to pathologies much earlier than arteries and veins.The capillary system functioning monitoring often makes it possible to detect disorders in vivo at an early stage, much earlier than when monitoring the condition of the veins and arteries.The quantitative parameters of microvessels can be not only a prognostic element, but also a measure of pathological processes and treatment effectiveness. 3,4e main function of the dermal bed (vascular system of the skin) is thermoregulation.However, the dermal vessels are also involved in wound healing, controlling inflammatory reactions, and maintaining tissue homeostasis.The rate of blood flow through the vessels is primarily influenced by the tension of the vessel walls.The endothelium influences the tension of the vascular wall by secreting nitric oxide (NO), prostacyclins, endothelin 1 (ET-1, endothelin-1), and endothelium-derived hyperpolarizing factor (EDHF). 5 The inflammation present during the acne lesions formation also affects exocytosis and increased vascular network.Inflammation leads to increased seepage of blood morphotic factors, including red blood cells. 6This causes erythema around acne lesions.We have made an attempt to assess the microcirculation of the skin within acne lesions and determine whether, based on the changes in microcirculation, it can be anticipated as to the intensity of the changes that arise, or to assess the effectiveness of treatment or the tendency to scarring.
The aim of the study was to assess the microcirculation of acne lesions using noninvasive skin bioengineering methods-semiquantitative blood measurement based on laser speckle contrast analysis (LASCA) and quantitative assessment of the severity of acne lesions with the use of hyperspectral imaging supported by methods of image analysis and processing.

| Patients
Twenty-five volunteers with diagnosed moderate acne (mean age

| LASCA method-perfusion assessment
The skin of the volunteers was assessed by a noninvasive semiquantitative blood perfusion measurement method based on LASCA using the Pericam PSI NR device, Perimed, Sweden.Analysis was performed based on averaging the blood flow results from 21 images/s for each acne lesion, leaving an effective refresh rate of 0.2 s/image.
The mean perfusion was calculated by PimSoft software version 1.5. in the points of inflammatory changes in the course of acne and healthy skin.LASCA, also known as laser speckle contrast imaging (LSCI), is a method of instantly visualizing blood microcirculation in tissues.LASCA works according to the following principles: when coherent laser radiation hits the surface of the skin, it is randomly scattered in the tissue, and produces a random interference pattern called a speckle pattern.The applied laser radiation is selectively absorbed by the erythrocyte hemoglobin.If the erythrocytes in the blood vessels move, the contrast within the speckle pattern will change as the movement of the scattering particles induces Doppler shifts in the laser light.This phenomenon can be semi-quantified by computational speckle contrast analysis and can provide information about the microcirculation of the blood flow (the speed of red blood cells traveling).The faster the erythrocytes move (higher perfusion), the more blurred the speckle pattern.On this basis, the speed of blood flow through the cutaneous vessels and, thus, the perfusion of the skin in perfusion units (perfusion index) can be determined. 7,8A total of 75 inflammatory acne lesions and 70 areas of skin not affected by acne lesions, understood as a control, were examined.The control areas were located in the same area, and no changes were clinically verified in this area by three dermatologists independently.Each examined area (acne and control) was selected arbitrarily by experienced dermatologists.

| Hyperspectral imaging
The tissue imaging with a hyperspectral camera shows the potential for noninvasive disease diagnosis and severity.Radiation delivered to biological tissue is diffused due to the heterogeneity of biological structures and is absorbed mainly in hemoglobin, melanin, and water as it propagates through the tissue.The tissue absorption and scattering characteristics change as the disease progresses, so total reflectance analysis shows quantitative diagnostic information about tissue pathology. 9The analysis was performed using a Specim IQ hyperspectral camera (Specim, Oulu, Finland).The images have a spatial resolution of 512 × 512 pixels, and a spectral resolution of 3 nm, which gives 204 bands in the range from 397 to 1004 nm.
Hyperspectral imaging is a technology for obtaining quantitative measurements from transcutaneous spatial and spectral information.The main advantage of hyperspectral imaging is to obtain hundreds of spectral bands containing data that is, a fusion of image and spectral data over a wide spectral range.The images obtained from the hyperspectral cubes can be used to assess the reflectance value in order to determine the properties of the tissue (including the content of epidermal and skin chromophores and their distribution in the studied space).An important advantage is that hyperspectral imaging is noninvasive and does not require patient preparation for examination. 10,11om the hyperspectral image saved in the *.dat format of acne skin, images corresponding to acne lesions were arbitrarily isolated and converted to the *.tiff format.It gave a total of 3060 photos.
The obtained images were then converted to 256 shades of gray for gray-level co-occurrence matrix analysis (GLCM).

| Gray-level co-occurrence matrix analysis
The acne lesion is the more visible, the higher the contrast between the healthy skin and the acne lesion.At the same time, the more changes on a unit of the skin surface, the more severe the acne.As per the described approach, an algorithm that quantifies this relationship, that is, the contrast between healthy skin and acne lesions, has been proposed for quantification.A dedicated version of the algorithm, based on the GCLM, was used.
GLCM determines the spatial relationship between pixels by quantifying the difference in intensity (contrast, homogeneity) between a test pixel and its adjacent pixels.Thus, the GLCM informs about how often the pixels of a given brightness are located at a certain distance from each other.Pixels that are adjacent to each other are analyzed the most often. 12CM analysis enables parameters such as contrast and homogeneity to be obtained.Contrast is the difference between the highest and lowest values of adjacent sets of pixels (if adjacent pixels are examined).High contrast is usually equated with rough textures, while low contrast is equated with smooth.For homogeneous images, the contrast value is 0 or close to 0. Homogeneity is a measure of the homogeneity of an image, and it ranges from 0-1.Higher homogeneity values show less differences in gray among the pixel set.
The homogeneity is maximal when all elements in the image are the same. 13Contrast (equation 1) and Homogeneity (equation 2) show an approximately inverse correlation with respect to an equivalent distribution in the population of pixel pairs.Thus, this means that the homogeneity decreases as the contrast increases, and vice versa. 14,15ntrast (1) Homogeneity (2)   where: i-brightness of the tested pixel, j-brightness of the adjacent pixel, p (i, j)-value of the weight matrix.

| Statistical analysis
The statistical analysis of the results was performed with the use of OriginPro and Statistica 13 software.Normality of the distribution was checked using the Shapiro-Wilk test and the quantile-quantile plot.The results at the level of p < 0.05 were considered statistically significant.Differences in perfusion, reflectance, contrast and homogeneity of acne-affected and healthy skin were tested using the t-student test for independent samples.

| Blood perfusion in inflammatory lesions in the acne skin
The study was performed on inflammatory pustular and follicular lesions in which increased perfusion was recorded.The increase in perfusion relative to the skin unaffected by lesions ranged from 88% to 162%. Figure 1 shows how the skin shows a higher blood perfusion in the area of inflammatory lesions compared to the untouched skin.The higher the perfusion, the higher the blood flow by the tested tissue.
A color scale represents the blood perfusion in the skin.The red color in the test area corresponds to a high perfusion unit (PU) value and the blue color to a low PU value.

� i,j
�i − j� 2 p(i, j) There was a statistically significant difference between blood perfusion at the site of the acne (inflammatory) lesion and healthy skin (Figure 1).On average, perfusion in the inflammatory lesion is 117% higher than in the skin unaffected by the acne lesion (Figure 2).

| Acne skin reflectance
The analysis of hyperspectral parameters allowed for the quantitative identification of the absorption/scattering of radiation in the range of 400-1000 nm in the examined area of the skin.The reflectance of acne lesions area and healthy skin were compared.Each pixel in the acquired hyperspectral image has a brightness value that corresponds to the skin reflectance.The brighter the pixel, the higher the reflectance of the skin in this area (reflects more radiation).In view of the above, thanks to the hyperspectral imaging method used in this way, it is possible to find the wavelength that corresponds to the highest contrast in a given examined area (acne lesion vs. healthy skin).This enables very precise, quantitative imaging of acne lesions.
Hyperspectral imaging made it possible to determine the wavelength (in the studied range) for which there is the greatest difference in reflectance between acne lesions and healthy skin.
The reflectance (Figures 3 and 4) of acne lesions is lower in the range of 400-600 nm compared to the skin not affected by acne lesions (control skin).
Three points of the highest decrease in reflectance were recorded: 434, 549, and 588 nm for acne lesions (Figure 5), that is, these are the wavelengths for which the difference in reflectance between acne lesions and healthy skin is the greatest.Determination of the wavelengths for which the difference in reflectance between the affected and healthy skin may be of significant diagnostic and prognostic value.A statistically significant decrease in the reflectance of acne lesions was demonstrated for the abovementioned wavelengths that is, 434, 549, and 588 nm (Figure 5).

| GLCM analysis
The contrast of acne skin images is higher than that of healthy skin (control).This is the result of the presence of a large number of acne lesions, which causes large differences in brightness between adjacent pixels.What is worth noticing is that the proposed GLCM scale is perfectly correlated with the visual scales since the visual acne analysis is also based on the assessment of the contrast between the affected and healthy skin.Importantly, the applied spectral range of the test almost entirely coincides with the spectrum of visible radiation (380-680 nm).Therefore, the following quantitative scale of acne lesions can be proposed: the higher the GLCM contrast, the F I G U R E 1 Visualization of microcirculation in the face acne skin (A) and a photograph taken in cross-polarized light (B).1-3-acne lesions, 4-healthy skin area.

F I G U R E 2
Average blood perfusion assessed on the skin of all volunteers at the site of acne lesions and on healthy skin.Average, standard deviation, min-max, p < 0.001.greater the severity of acne.Due to the mathematical parameters of the algorithm used, this correlation is linear.
There was also a statistically significant difference in the contrast between the groups for 434 nm (p < 0.01), 549 nm (p < 0.001), and 588 nm (p < 0.001).One can observe an increase in contrast around 582-591 nm and a bulging of the plot (Figure 6), which means the highest irregularity of the structure and low reflectance.The homogeneity values are inversely proportional to the contrast value.
The higher the homogeneity, the more structurally homogeneous the image becomes.(Figure 7).

| DISCUSS ION
Currently, in clinical diagnostics, including acne vulgaris, dermatologists mainly recognize and evaluate skin features with the naked eye using qualitative scales. 6However, the assessment with the naked eye is largely subjective and depends on the experience of the doctor (or other specialists), color perception, and sensitivity of the human eye, the conditions of observation (the amount of light, color temperature of light, the angle of light), and even the emotional state of the patient (stress can induce skin erythema). 7Despite concerns about these factors, objective analysis is sometimes required in clinical assessments such as quantitative diagnosis and prognosis.The assessment of acne lesions is currently carried out primarily based on of qualitative methods.Visually, the number and advancement of changes are assessed based on of scales such as the Investigator's Global Assessment or Global Acne Grading System.Nevertheless, there are no scales that would allow the quantification of acne lesions.We are looking for tools that would allow for the quantitative assessment of skin lesions.Such tools are required to: be biometric (allow for quantitative analysis), be easy to use, not require the use of expensive instruments, allow for objective analysis regardless of external conditions, not be time-consuming, and enable the comparison of results between different centers.The visual difference between healthy skin and acne-altered skin is the effect of inflammation.Inflammation induces lymphocytic infiltration and increases the expression of inflammatory vascular markers that promote angiogenesis. 16The conducted research clearly shows a significant increase (two times on average) in skin perfusion in the area of acne lesions.Increased blood perfusion in acne inflammatory lesions compared to healthy skin suggests increased chemotaxis of inflammatory and wound healing cells. 16,17Additionally, cells are supplied with more oxygen and nutrients due to accelerated metabolism and inflammatory infiltration.and abnormal differentiation in unoccupied follicles that were similar to normal control follicles. 18Hence, increased perfusion may suggest an inflammatory lesion in the course of acne that is, not clinically exposed.At the same time, the obtained results may suggest that the use of anti-inflammatory substances (which has been clinically confirmed for years) and anti-angiogenesis substances (which is new) may inhibit the formation of acne lesions.

At the molecular level, an increase in the expression of integrin
Research by Holland et al. 19 showed an increase in the number of blood vessels in acne lesions in patients without and with a tendency to scarring.A difference was observed in EN4 + labeled blood vessels in prone to scarring and non-susceptible individuals.By 72 h after of inflammation, blood vessel counts had returned to normal in scar-prone patients, but remained significantly higher in these patients than in non-scarred patients.Evolving inflammatory lesions showed significant angiogenesis and expression of vascular adhesion molecules with preferential recruitment of large numbers of CD4 + T cells and influx of macrophages and Langerhans cells.The formation of acne scars is associated with persistent inflammation, as a result of which there is an atrophogenic effect in the connective tissue F I G U R E 4 Reflectance of skin ROI: acne lesion and control skin in the wavelength range 400-1000 nm (A).Reflectance of acne lesion and control skin in the wavelength range 400-600 nm in the scale unit 0.02 for a more accurate image (B).Example based on single results (acne and control).

F I G U R E 5
The reflectance of acne lesions and healthy skin for the wavelengths of 434 nm (p < 0.001), 549 nm (p < 0.001), and 588 nm (p < 0.001).
(atrophic scars) or hyperproliferative changes in fibroblasts (hypertrophic scars).Acne treatment should also target measures that influence the regulation of immune mechanisms and angiogenesis.
Results of acne lesion development imaging using OCT-based microangiography by Baran et al. 20 show the stages of development of an acne lesion that damages the microvessel network in the inflammatory stage, which is caused by the formation of puffiness inside the epidermal layer, which in turn blocks or damages the microvessels in the dermis.The skin imbalance naturally leads to the hypertrophic scarring seen with dense microvascularization.At the end of the acne healing process, the microcirculation network becomes less dense than in normal tissue, but fibrosis can be observed.Therefore, the proposed methods of perfusion assessment may be useful in clinical trials, for example, of anti-scarring or antiinflammatory substances.
Increased blood perfusion from acne lesions is the result of a significant increase in the amount of hemoglobin.Therefore, one may expect that laser techniques and other optical techniques based on selective photothermolysis may also have clinical significance in the treatment of acne.What is worth noting is that laser radiation selectively absorbed by hemoglobin, which will reduce the number of vessels, may, however, induce other mechanisms of acne development-for example, by stimulating the secretion of some proinflammatory cytokines. 21reover, the perfusion value can reflect the blood vessel density and the severity of inflammation in the acne lesion, so it can be used as a biomarker for the quantification of acne severity and the biometric evaluation of treatment efficacy.
The conducted research shows that the lowest reflectance was demonstrated for the three wavelengths: 434 nm, 549 nm, and 588 nm.Importantly, these wavelengths resemble the peaks of the maximum absorption by hemoglobin. 22e wavelength for which a significant decrease in reflectance is observed is 434 nm radiation of this length is absorbed more intensively.Behind the absorption of radiation with a wavelength of 434 nm, there is probably coproporphyrin III produced by Cutibacterium acnes, because it absorbs blue light in the 400-450 nm range to the greatest extent, although the absorption peak is 415 nm. 23,24Coproporphyrin III is a porphyrin group molecule that exhibits a characteristic pattern of optical porphyrin absorption.It consists of the Soret band, the major absorption peak around 400 nm, and the Q-bands, the four weaker absorption peaks, which lie in the range 450-700 nm. 25 This range is in line with the lowest reflectance results in our study.Confirmation of such a correlation in subsequent studies may contribute to an attempt to quantify skin colonization by C. acnes using hyperspectral imaging.The common use of the reflectance/absorbance assessment method (e.g., UV/UV-VIS spectroscopy) allows the identification of spectral parameters of only a small area-not exceeding 1mm 2 .The applied hyperspectral imaging method enables simultaneous measurement of the entire skin of the face and neckline with a satisfactory resolution-in this case 512 x 512 pixels.
As indicated above, the quantification of radiation absorption in the wavelength range corresponding to the absorption maximum for the chromophores produced by C. acnes may be an important aspect in terms of the degree of colonization of acne lesions by these bacteria.Considering that one of the main strategies of acne therapy is antibiotic therapy-the degree of colonization of acne lesionsexpressed as the absorption value for coproporphyrin III-may be of significant importance in the context of optimizing antibiotic therapy, assessing the effectiveness of treatment and developing new methods of therapy based on light therapy and laser therapy.This suggests new therapeutic possibilities with the use of for example, high-energy light with a wavelength in the range of 400-599 nm, due to the lower reflectance in comparison to the range of 600-1000 nm.The proposed hyperspectral imaging method also allows to determine for which wavelengths the greatest difference in reflectance between the affected and healthy skin is observed.Thus, it is possible to optimize optical methods, including laser ones, so that the length of the applied therapeutic radiation corresponds to the minimum reflectance.This is important because only the radiation absorbed by the skin leads to specific biological effects.At the same time, determining the maximum difference between the reflectance acne-affected skin and a healthy one allows laser radiation (or other radiation) to be optimized so that the absorption of radiation by acne lesions is maximum (maximizing the therapeutic effect) and at the same time minimizing through the healthy skin (minimizing side effects).In the adopted research model, the optimal wavelengths will be 434 nm, 549 nm, and 588 nm.Of course, it should also be taken into account that as the wavelength of radiation decreases, it has a lower ability to penetrate the skin.Therefore, it may not reach the target chromophores.The radiation with a wavelength of 588 nm seems to be optimal in acne laser therapy (taking into account strictly the reflectance and penetration parameters).
Hyperspectral images of acne lesions show higher contrast values and lower homogeneity as compared to the skin without pathology.
This indicates an increase in the skin's chromophore content and an uneven distribution, possibly due to inflammation.
The conducted studies are the first to quantify acne perfusion using the LASCA method.They jump to a significant, about twofold, increase in perfusion in the area of acne lesions.At the same time, the proposed methods of hyperspectral imaging allow for the optimization of the interaction of light radiation with the skin, including acne lesions, and the proposed methods of contrast assessment (GLCM) allow for a precise, quantitative assessment of acne lesions.

23 ± 4
years) and the control group-20 people without acne lesions on the face (mean age 24 ± 6 years) were qualified for the pilot study.Each volunteer provided a written consent in accordance with the Declaration of Helsinki.Volunteers were qualified by three experienced dermatologists.Over 3000 images of acne lesions were recorded and examined within both groups.Criteria for inclusion in the research: • Male and female patients who are 18 years of age and under 40 years of age, who can independently decide about participation in the study • Determining the presence of inflammatory acne lesions • Increased secretion of sebum • Increased keratinization of the mouths of the horny follicles or the stratum corneum of the epidermis • Changes related to the incorrect distribution of the pigment in the skin • Informed consent to participate in the study Criteria for exclusion from research: • Age under 18 and over 40 years of age • dermatological treatment in progress or completed not earlier than 3 months before qualifying for the study • Pregnancy and breastfeeding • Acquired and congenital albinism • Seborrheic dermatitis • Active viral (e.g., herpes), parasitic and fungal infections • Diagnosed: systemic lupus erythematosus, demyelinating diseases, active neoplastic disease, mental illness • Absence or withdrawal of consent to participate in the study The volunteers were asked to clean their face with a cleansing gel 3 h before the test and not to use other cosmetic preparations for the following 24 h.The reason for this was to simulate the physiological condition of the skin and to avoid light waves being reflected or absorbed by active substances in cosmetics.

α
v in the skin vasculature is a signal for the initiation of angiogenesis and an increase in the number of vascular adhesion molecules, intracellular vascular adhesion molecule 1 (VCAM-1), E-selectin, and intercellular adhesion molecule 1 (ICAM-1), which facilitate the recruitment and migration of inflammatory cells to the dermis.No neutrophils were observed in the infiltrate, indicating initiation of inflammation by CD4 T cells in response to antigenic stimulation.Moreover, inflammation occurs in the absence of hyperproliferation F I G U R E 3 Nine subchannel images captured by the hyperspectral camera from a volunteer acne skin converted to grayscale.

F I G U R E 6
Contrast of hyperspectral images for acne lesions and healthy skin for individual wavelengths.F I G U R E 7 Homogeneity of the hyperspectral images for acne lesions and healthy skin for individual wavelengths.