1. Introduction

Noninvasively detailed visualisation of the external and internal structures of tissue is always the key to perform a comprehensive examination and make an accurate diagnosis [1–3]. Biomedical imaging technology plays an indispensable role in modern medical systems [4,5]. Nevertheless, despite many years of development, it remains a challenge to visualise the internal fine structure of skin tissue with high spatial resolution, deep penetration, and strong contrast in a non-invasive manner [6–8]. Optical coherence tomography (OCT) has been used for label-free imaging of skin tissue constructs with a relatively high resolution [9], but the imaging depth still limits to the high optical scattering of biological tissue [10]. Moreover, OCT angiography (OCTA) utilizes the intrinsic motion contrast to differentiate functional blood vessels from static tissue background [11], but static structures (for example, stratum basale and a condition in which blood vessels have no flow of red blood cells) of skin tissue cannot be visualized. Ultrasonography can penetrate skin tissue up to several centimeters [5,12], but it is limited by a low contrast for distinguishing microvascular structures with dozens of micrometres beneath skin surface [13]. Biopsy is commonly used to evaluate and analyse skin diseases [14]. However, the technique usually accompanies with invasiveness and will damage surrounding healthy skin, so that it cannot be a timely and repeatedly screening method in the in clinical practice. In view of the limitations of current diagnostic techniques, clinicians face great difficulties in making accurate classifications, optimising therapeutic regimens, and quantifying and evaluating therapies for skin diseases, such as port wine stains (PWS) – A kind of congenital and non-self-healing dermal vascular malformations with a high incidence of 3‰-5‰ [15], gradually expands with age and deteriorates into deep tissue infiltration. PWS can result in Sturge-Weber syndrome, which will cause tissue calcification and nerve injury in cerebral cortex, leading to disability and death. Since 1980s, cryotherapy, intense pulsed laser therapy and photodynamic therapy have been used in the clinical treatment of this disease, but so far, the complete cure rate of PWS is still less than 20% [16], which mainly lies in the inability to non-invasively identify the microvessels (tens of microns in diameter) in the lesion dermis in-situ, and then to accurately control the dose and duration of treatment, often leading to longtime treatment or ineffective treatment.

The recent development of photoacoustic microscopy (PAM) combines optical and ultrasonic merits to reduce the tissue scattering of photons with one-way ultrasound detection while retaining the high optical contrast [17–20]. PAM emerges as a powerful imaging modality in the mesoscopic range, which provides high-resolution structural insights on the skin in vivo [21,22]. As an optical-contrast-based imaging technique, the photoacoustic (PA) detects endogenous skin chromophores [23,24], i.e., melanin and haemoglobin. The complex structure of human skin produces broadband PA signals (from a few to hundreds of megahertz) owing to the wide variability of the absorbers’ size, including the pigment granules and the vasculature with a few to hundreds of micrometers [25,26]. Therefore, the spatial-scale of excitation beam and the bandwidth of acoustic detection should be trade off according to the size and depth of the target [26–29]. In terms of the pathological features and the clinical demand of skin diseases, in this paper, we developed a noninvasive photoacoustic microscopic biopsy (PAMB) system attached with a featured photoacoustic index and a clinical application scheme by using an adjustable confocal opto-sono objective. The opto-sono objective is mainly composed of custom-made switchable objective lenses and a focused broadband (5–90 MHz) transducer. The imaging strategy is the use of switchable objective lenses with different numerical apertures (NAs), as well as a multiscale adjustable configuration (adjustable range of focus: 0–3 cm) to regulate the optical-acoustic focus under the skin surface. By switching a high NA objective lens, the PAMB can ensure high resolution to resolve most of the elements within epidermis. Alternatively, by selecting appropriate NA objective lens and optical-acoustic focusing depth under the skin surface, it allows for more photons to scatter into deeper skin tissues to visualize parts of the dermis and sub-dermal layers. The optimal combination of NA modes, focusing depths, and frequency equalization therefore can bring new insights to image multiple layers of skin. The PAMB provides a smooth transition from optical resolution mode in microscopic imaging of superficial skin layers to acoustic resolution mode when imaging at greater depths within intensely scattering deep skin layers.

2. Experiment and methods

2.1 PAMB system

The system is based on an adjustable confocal opto-sono objective (Fig. 1 and Supplement 1 Fig. S1), constructed to yield a lateral resolution of 1.5 µm and an axial resolution of 34 µm to depths of several millimetres (Supplement 1 Fig. S2). Typically, lateral resolution measurements were performed at 1.5–3.8 µm (zero plane), 10–40 µm (0.2–1.5 mm depth in skin tissue), and 40–104 µm (1.5–3.0 mm depth in skin tissue). Here, a lateral resolution of 10–40 µm is important for distinguishing vascular features of PWS skin between imaging depths of 0.2 mm and 1.5 mm. The opto-sono objective scans perpendicular to its axial axis and is perpendicular to the tested surface. The optical-acoustic confocal focus is maintained slightly under the tested surface of the skin.

 

Fig. 1. Noninvasive photoacoustic microscopic biopsy (PAMB) system using an adjustable confocal opto-sono objective. (a) Schematic of optical excitation and photoacoustic (PA) signal generation by multilayered structures of Asiatic skin. SC, stratum corneum; SB, stratum basale; EDJ, epidermal-dermal junction; SVP, superficial vascular plexus; DVP, deep vascular plexus. (b) Schematic of opto-sono objective. SMF, single-mode fiber; FC, fibre collimator; BE, beam expander; SOL, switchable objective lenses; MAD, multiscale adjustable device; UT, ultrasound transducer. (c) Resolution of PAMB at focus depth position in phantom. The depth/resolution ratio represents imaging performance of the PAMB system, and the larger the value, the better the imaging performance of the system.

Download Full Size | PDF

A miniature laser (Model DTL-314QT, Pulsed Q-switched Lasers, Russia) operating at a 4-ns pulse width at 532-nm with a 10-kHz repetition rate was used as the radiation source and provided sufficient contrast for PA imaging. The laser is focused by a convex lens, which passes through a 25-μm pinhole for spatial filtering and then coupled into a single-mode fibre using a plan objective lens [NA of 0.1; working distance (WD) of 37.5 mm]. The laser is finally focused by the adjustable confocal opto-sono objective and irradiates the test area to produce PA signals. In the integrated scanning head, the opto-sono objective is automatically actuated by a two-dimensional scanner (scan distance of 5 × 5 cm2, scan resolution of 0.1 µm, HRXWJ-50R-2, Beijing TianRuiZhongHai Precision instrument Co., Ltd, China), which was used as a mechanical scanning platform. The excitation source is eventually focused into switchable NA modes, typically 4×/0.1, 10×/0.3, and 20×/0.55. An articulated arm helps to fix the casing to the imaged area, minimising motion artefacts in the reconstructed images. A transparent plastic film is fixed to the head of the coupling cup for the penetration of light and sound. The coupling cup is filled with 5 ml of deionised water and enables acoustic coupling between the tested surface and detection surface of the PVDF transducer (Supplement 1 Fig. S1).

Each PWS skin B-Scan image was generated using PA signals collected over 0.5–5 mm (50–500 scan points) and within 5–50 ms, and the separation distance between B-Scan images was 10 μm. The duration of the entire process for preparing the PWS volunteers, including placement of the scanning head on the tissue surface, was 3–5 min. The collected PA signals were sequentially pre-amplified with a 50-dB low-noise amplifier (LNA-650, RF Bay Inc., USA), digitised using a dual-channel data acquisition card (M4i.4450-x8, Spectrum, German), and stored on the computer’s hard drive for subsequent data processing.PAMB conforms to the human-use safety limits described in the American National Standard for Safe Use of Lasers (ANSI Z136.1-2014, Laser Institute of America; Methods: Safety limits).

2.2 Efficacy assessment

An independent group of three experienced pathologists who were not involved in this study visually reviewed the photos and experimental PAMB images and assessed the patients’ condition and treatment efficacy. The pathologists made their assessments independently. The results were deemed valid only if two or more pathologists agreed. If no agreement was reached, the condition and treatment response were re-evaluated until consensus was achieved.

2.3 Image reconstruction and depth compensation

Data were collected using integrated scanning head acquisition assuming an x-y plane parallel to the tested skin surface. At each position, several light-absorption mode lines (‘A-lines’) were obtained along the z direction.

In PAMB, the attenuation of PA signals f(t) is mainly owing to optical attenuation and acoustic attenuation. Therefore, the integrated optical and acoustic attenuation coefficient, which represents the PA signal depth compensation F(t), can be expressed as [30–32]:

After the PA signals pass subsequent wavelet denoising, Wiener filtering, and depth compensation, the original sound-pressure distribution of the single position P(t) can be obtained. Then, the relative optical absorption distribution of the entire scanning region can be obtained by projecting the original sound pressure distribution of multiple positions. Finally, reconstruction of the optical absorption image is achieved by obtaining the maximum intensity projections of the reconstructed images along the direction shown.

The image quality of PAMB was improved by using depth compensation algorithms, which effectively solved image distortion owing to the extreme attenuation of high-frequency PA signals in the z direction (depth).

2.4 Volunteer selection, skin imaging, data grouping and general statistics

All volunteers were imaged following approval from the Ethics Committees of the First medical center of PLA General Hospital, Beijing, China (Clinical Trials. gov number, 2017 Ethic Review No. 012) and the Nanfang Hospital of Southern Medical University (Clinical Trials. gov number, NFEC-2017-093). Before skin imaging, verbal and written informed consent was obtained from all adult volunteers, and parental consent was obtained for volunteers younger than 18 years. Consent for publishing skin photos was also obtained (Figs. 2–7 and Supplement 1 Figs. S3–S5). Patients’ conditions were evaluated by the three experienced pathologists to confirm the clinical diagnosis.

There were two groups of volunteers. The first consisted of one scar patient, and three healthy individuals; the data obtained from this group were used to evaluate the effect of scalable adjustments on PAMB of the skin. The second group consisted of 174 patients with PWS. Among them, the number of patients with pink lesions, purple lesions, and proliferative lesions was 62, 67, and 45. The number of patients aged 5 to 15, 16 to 30, and 31 to 50 was 69, 63, and 42, respectively. At least six B-Scan PA images were taken at each test-area of the volunteers, and the separation distance between each B-Scan PA image was 10 μm. The data obtained from this group were used to evaluate the performance of PAMB in the calculation and evaluation of vascular features of PWS (Figs. 3–7 and Supplement 1 Figs. S3–S5).

Paired t-tests and Wilcoxon signed-rank tests were used to assess the significance of significant differences in the metrics used to compare different skin colours, different stages of PWS, and before and after PDT treatment of PWS. The results of the colourimetric evaluation and PWS vascular feature parameters were compared using Pearson correlation coefficients. P ≤ 0.05 was considered statistically significant.

2.5 Quantification of vascular features in patients with PWS

The epidermis is located at the top of the skin, and can be used as a starting point for the PAMB depth reference. Here, the average vascular depth was defined:

The average vascular diameter can be obtained by the following formula:

The relative vascular area in the B-Scan PA images can be obtained by the following formula:

Here, ${A_y}(x,z) = \left\{ {\begin{array}{c} {0,x < \varepsilon }\\ {1,x \ge \varepsilon } \end{array}} \right.$, where ε is the pixel value used to distinguish between the vasculature and background.

A statistical analysis of the vascular patterns is presented in Figs. 3–7 and Supplement 1 Figs. S3-S5. The lesion vessel depth and density are selected here as the main descriptive parameters because they have been widely accepted to describe and assess vascular morphology. Lesion vessel density (ρ) is defined as the ratio of the blood vessel area (S_p) to the total area of the tissue slice, represented by the B-scan structure area (S_b). First, all B-scan images were converted into binary images by setting a threshold value of 15 dB to eliminate the noise effect on the calculations. Sp and Sb were then determined by the pixel numbers of the binary image of the corresponding structural image. Finally, the value of ρ was calculated using the following equation [33,34]:

To further investigate the clinical relevance of PAMB-resolved features, the following severity index was formulated:

The depth, diameter and density of the elongated and dilated capillary loops were only calculated for PWS skin. All B-Scan PA images were first marked using two white dashed lines, which separated the epidermis and dermis. The images were subsequently processed using Image J. Subsequently, the diameter, depth and density of the lesion vessels were compared using nonparametric Wilcoxon matched-pair signed-rank tests and Pearson correlation coefficients. The vascular feature evaluations (blanching rate) before and after treatment were also compared using Wilcoxon matched-pair signed-rank nonparametric tests (Figs. 6(c)–(d)).

Photos of the volunteers were taken with a 24.2-megapixel digital camera (ILCE-6300L, Sony Corporation, Japan) illuminated using the on-camera flash. Each of the lesions was labelled with a specially designed label that contained a red marker and was graded on a scale of 0–9 [35,36]. All pictures were taken from a similar angle and distance, capturing most of the PWS lesion areas with no surface glare.

3. Results

3.1 Evaluation of the PAMB system and test case description

A high-resolution PAMB system was developed for immediate clinical application (Fig. 1 and Supplement 1 Fig. S1). PAMB can rotate to accommodate different detection positions when the integrated scanning head is assembled on an articulated arm, which couples conveniently to the small area of the skin surface under investigation. The backward coaxial confocal mode is beneficial to attain high signal-to-noise ratio and detection sensitivity, and the use of a broadband Polyvinylidene Fluoride (PVDF) transducer effectively avoids artefacts induced by limited detection coverage [37,38]. Figure 1(a) summarizes the protocol in which is proposed to evaluate skin conditions by PAMB in clinical practice, and illustrates PA signal generation in different skin layers. The excitation source is eventually focused on the surface of human skin using the objective lenses with three NA modes, typically 4×/0.1, 10×/0.3, and 20×/0.55 (Fig. 1(b)), respectively. Figure 1(c) graphically illustrates the correlation between the imaging depth and spatial resolution of PAMB. The acoustic nature of PA imaging allows it to easily achieve a high depth-to-resolution ratio (blue dashed line) of roughly 170, constant across three NA modes of PAMB. Resolutions and depth experiments were performed to verify that PAMB exhibits a lateral resolution of 1.5 μm, an axial resolution of 34 μm in focal zone, and an imaging depth of ∼3 mm (Supplement 1 Fig. S2).

To qualitatively evaluate the skin-imaging performance of PAMB (Methods), the B-Scan images of one palm skin were successively obtained at three NA modes (4×, 10×, and 20×), different focusing depths and different frequency bands, which allows for the rendering of fine spatial details together with switchable-resolution skin structures (Fig. 2). Then the PAMB was used to image skin structures at different depth scales from epidermis to subcutaneous tissue (Table 1), which is in accordance with radius and depth of the absorbers within each layer as reported previously [25,26,31,35,36,39–41].

 

Fig. 2. Comparison of in vivo contrasts from palm skin, opisthenar skin, and scar skin by PAMB. PAMB images of (a) healthy palm skin; (b) healthy opisthenar skin; and (c) scar skin from one Asian. PA images [from left to right: volume-rendered images of SC and SB, MAP images of dermal vascular (DV)] depicted along direction perpendicular to skin surface within limits marked in B-Scan sections. Different biological structures appear in different colour depending on spatial depth distribution (SC, SB, and DV appear mainly in yellow, blue, and red, respectively). SB can be distinguished from SC and EDJ, whose structures follow shape of epidermal ridges. Below SB, high-contrast DV structure is resolved (gradient colour varies with depth, which is called depth-encode). Note that DV consists of SVP (210–650 μm below surface) and DVP (650–1500 μm below surface) as shown in B-Scan sections. PAMB visualizes scar skin of relatively thinner epidermis but more dense horizontal plexus in dermis. Color bars are positively correlated with the PA signal amplitude.

Download Full Size | PDF

Table 1. Diameters of main absorbers, depths of layers, focusing depth, and NA modes for healthy human skin were compiled from experimental data by PAMB system

View Table | View all tables in this article

PAMB skin images (Fig. 2) revealed the stratum corneum (SC), stratum basale (SB), and dermal vascular (DV) of healthy palm skin (Fig. 2(a)), healthy opisthenar skin (Fig. 2(b)), and scar skin (Fig. 2(c)) of one Asian subject. The PAMB images depicted the superficial skin ridges of the SC in yellow, SB in blue following the shape of the epidermal ridges, and a maximum amplitude projection (MAP) of the DV in the vascular plexus. It can be easily determined that healthy palm skin has relatively thinner SB but a denser horizontal plexus in the dermis compared with healthy opisthenar skin. In addition, the SC and SB thickness of scar skin is thinner than that of healthy opisthenar skin, but the vessel density is higher compared with healthy skin. Consequently, PAMB has the ability to visualise variations in the skin layers, dermal vasculature, and capillary loops at a level of detail by performing scalable imaging for different skin conditions.

Clinical PA images (Figs. 3–7 and Supplement 1 Figs. S3–S5) clearly show the dermal vasculature of PWS skin, and the detailed clinical PA data on 174 patients is used for accurate assessment of PWS (Fig. 7 and Table 4). The same area of volunteer was repeatedly measured by PAMB to demonstrate that no significant changes in the vascular structure of the skin were induced via photothermal effects. This system provided images that showed structures at a depth of ∼1.5 mm for 532-nm illumination. The results show that PAMB can be used for the in vivo recognition of cells, superficial structures and deep subcutaneous blood vessels at a level of detail and resolution/depth ratio not achieved by other modalities. Therefore, the PAMB can offer never-before-documented scalable imaging capabilities for skin imaging by adjusting the operating mode of the opto-sono objective.

3.2 PAMB of PWS: calculation of dermal vascular parameters, and their clinical relevance

To explore whether PAMB can be used to quantitatively evaluate PWS lesions, three parameters used as clinically related indicators were computed (i) the diameter of lesion vessels, (ii) the depth of lesion vessels, and (iii) the density of lesion vessels. PWS lesion vessels are mainly distributed in the papillary and mid-reticular layers of the dermis, and typically have a diameter range of 30-300 µm [15]. To achieve sufficient penetration with good resolution of dilated lesion vessels, the 4×/0.1 NA mode was used to image PWS skin during clinical trials. The schematic of pathological characteristics for PWS skin is summarized, shown in Fig. 3(a), which indicates that there is a large increase in blood vessels of the dermis from the diseased area [15,16]. Then the PWS skin from one Asiatic PWS patient were imaged, revealing marked changes (dermal vasculature had greater diameter, higher density and deeper depth in PWS skin) compared with the healthy skin from the same patient (Fig. 3(b)–(e)). These findings are consistent with the histology of PWS skin [15,16].

 

Fig. 3. PAMB of healthy skin vs. symmetric port wine stain (PWS) skin. (a) Schematic of pathological characteristics for PWS skin. (b–c) PA images of upper jaw (healthy) skin and under jaw (PWS) skin of one Asiatic patient with PWS and photographs of corresponding detecting areas. (d–e) PA images of upper lip (healthy) skin and under lip (PWS) skin of the same patient and photographs of corresponding detecting areas. PA images (MAP images of DV, 3D PA dermal volumetric images and B-Scan sections) depicted along direction perpendicular to skin surface. Epidermis (EP) is denoted by two white dotted lines. Below EP, a dilated and dense vascular structure of dermis (DR) is resolved (gradient colour varies with depth).

Download Full Size | PDF

The next step was to investigate whether PAMB could be used to resolve and quantify the pathophysiological features of PWS skin. The diameter, depth and density of lesion vessels are the main parameters for assessing the level of capillary abnormality in PWS skin [14,15], and both were computed as the distribution scope and number of voxels in the dermis that corresponded to the inner lumen of blood vessels based on image segmentation (Methods). The depth and density of blood vessels in healthy skin and PWS skin were statistically assessed from the same Asiatic PWS patient (Fig. 4(a)). The results revealed significant differences (P < 0.01) in the depth and density of blood vessels between healthy skin and PWS skin.

 

Fig. 4. Calculation of dermal vascular parameters of PWS. (a) Statistics, i.e., feature parameters of lesion vessels from the PA images of patient (**p < 0.01; mean ± s.d.). (b) PAFIND index vs. abnormal lesion grade (EI) from experimental images of Asiatic PWS patients (n = 10 per group). Blue line represents ideal case in which PAFIND index corresponds exactly to EI. (c–f) Use of area under the ROC curve (AUC) and confidence intervals (95% CI) to distinguish between healthy skin and PWS skin from 45 PWS patients, as calculated using the parameters (diameter, depth, density, and PAFIND index) of blood vessels (BVs) for n = 90 independent skin regions (n = 45 healthy skin / n = 45 PWS skin), are shown.

Download Full Size | PDF

Finally, the clinical relevance of PAMB features were evaluated by using the PAFIND (Index of PA features), a function of microvessel density and depth, which can be used to quantitatively summarise the features and severity of PWS skin (Methods). The PAFIND is a comprehensive parameter and can effectively reflect the microvessel density and depth of PWS skin, which should be an accurate and effective parameter in clinic. To compare the abnormal grades of different PWS patients, we graded each lesion on a scale of 0–9 using the EI image analysis [42,43]. Significant differences (p <0.05) were identified among the EI grades of PWS skins as assessed by paired t tests in experimental PA images. By the correlation between PAFIND and EI, the PWS skins can be evaluated that the degrees of abnormality before and after treatments are well graded, as shown in Fig. 4(b). In addition, forty-five Asiatic PWS patients [In total n = 90 scans from independent skin regions of the patients (n = 45 healthy skin / n = 45 PWS skin) were acquired] were selected to obtain PA images by PAMB system. The ROC analysis revealed excellent ability of PAMB-derived dermal blood vessels to distinguish healthy skin from PWS skin (Table 2 and Figs. 4(c)–(f)). Thus, the PAFIND is a comprehensive parameter that can effectively reflect the degrees of abnormality in different PWS patients and should be an accurate and effective parameter in clinic.

Table 2. ROC analysis for PAMB-derived dermal blood vessels from PWS skin

View Table | View all tables in this article

3.3 Quantitative evaluation of photodynamic therapy (PDT) of PWS using PAFIND

The colour of PWS skin is often used in evaluation of the treatment efficacy, but it only reflects the blood vessels in the superficial skin (∼200 µm) and cannot be used to evaluate the whole diseased vascular region of PWS skin [16]. The key factor determining the treatment efficacy is the vascular structure (dilated vascular diameter, density and depth) of PWS skin [44]. Therefore, PWS vascular condition obtained by PAMB is a new scheme for treatment evaluation.

The next goal of the study was to investigate whether PAMB could be used to evaluate the curative effect of PWS skin after PDT [44,45]. The PAMB images before and after PDT treatment were examined to display morphological skin alterations, capillary loop elongation, and dermal vasculature changes. A PWS region measuring 3 mm × 3 mm from one Asiatic patient was imaged before and after two sessions of PDT (Fig. 5(a)-(b)) and showed marked changes. The results indicated that the diameter, depth, and density of lesion vessels of the woman in Fig. 5(a)-(b) were markedly reduced, as shown in Fig. 6(a). It also can be seen that the microvessels in the superficial dermis decreased significantly, but the larger vessels in the deep dermis remained hardly changed. In addition, significant differences in the vascular morphology of PWS skin were noted after PDT treatment (P = 0.004, 0.007 and 0.009 for the density, depth, and diameter of lesion vessels, respectively).

 

Fig. 5. Evaluating curative effect by PAMB in Asiatic patients with PWS before and after photodynamic therapy (PDT). (a)-(b) PA images and photographs of PWS skin before and after two sessions of PDT treatment in a 32-year-old Asiatic woman.

Download Full Size | PDF

By the correlation between PAFIND and EI, and the degree of PWS skin before and after PDT treatment are well graded using PAFIND, as shown in Fig. 6(b). Consequently, the experimental results show that PAMB can vividly visualize the full lesion profile and simultaneously quantify the dermal vascularisation, which indicates that it will be a valuable tool for helping clinicians to quantitatively evaluate pathological types and monitor the effects of treatment. The correlations between the PWS vasculature parameters (diameter and depth) and the bleaching rate were analysed, which are typically indicative of skin colour recovery and vascular morphology reconstruction, respectively [14]. Five Asiatic PWS patients (seven B-Scan images of PWS skin from each patient, Supplement 1 Fig. S5) were selected to obtain images before and after PDT treatment (Table 3), and analysed the diameter and depth of lesion vessels before and after PDT treatment. The correlation analysis reveals that the Pearson correlation coefficients are -0.406 and -0.413 for the diameter and depth of lesion vessels associated with the efficacy outcome, respectively. These results indicate that the differences were statistically significant (P < 0.05). Both the depth and diameter of PWS blood vessels were negatively related to the efficacy outcome of the PDT treatment (Figs. 6(c)–(d)). This finding is an example of how PAMB can be employed to evaluate the treatment of skin diseases.

 

Fig. 6. Calculation of dermal vascular parameters of PWS before and after PDT treatment. (a) Lesion vessel (LV) statistics from PA images of detection areas in Fig. 5 (**p < 0.01; mean ± s.d.). (b) Difference in abnormal lesion grade (EI) between measurements in pre-treatment and post-treatment PWS skin by using PAFIND index. Orange and grey portions correspond to values from median to third quartile and values from median to first quartile, respectively. (c) Blanching rate changes with vascular diameter. Pearson correlation coefficient of vascular diameter groups was -0.406. Difference was statistically significant (p = 0.035) and suggests that vascular diameter was negatively related to efficacy outcome of PDT. (d) Blanching rate changes with vascular depth. Pearson correlation coefficient of vascular depth groups was -0.413. Difference was statistically significant (p = 0.039) and suggests that vascular depth was negatively related to efficacy outcome of PDT treatment. Epidermis (EP) is denoted by white dotted lines.

Download Full Size | PDF

Table 3. Response of PWS to PDT

View Table | View all tables in this article

To further show that diagnostic features can be computed quantitatively from PAMB images, the differences in PWS data between the face and lip, central facial skin and lateral facial skin (Figs. 7(a)–(b) and Supplement 1 Fig. S3) were analysed. The results indicate that the variation in histologic vascular features, may explain the differences in therapeutic outcomes between different PWS skin positions. In particular, these data suggest an explanation for the clinical phenomenon wherein central PWS respond poorly to PDT compared with lateral PWS of the face [14]. The affected areas on the skin of the patients with PWS at different stages (pink, purple, and proliferative; Table 4 and Supplement 1 Fig. S4) were also imaged. The results revealed significant differences (P < 0.01) in the vessel depth and density between PWS lesions at different stages (Fig. 7(c)). The differences in PWS data between various age groups indicated that the degree of abnormality in patients correlated positively with age (Fig. 7(d)). Then the morphological features (average density, average diameter, maximum diameter, average depth, and maximum depth) of dermal vasculature between PWS skin and healthy skin from the 174 patients were compared by PAMB system respectively (Table 4), which was used to evaluate the effectiveness and reliability of the PAMB. These results further support the accuracy of the information that can be extracted from PAMB images, which potentially enable the accurate PAM-based near-histologic assessment of skin diseases.

 

Fig. 7. Histogram distribution for index of photoacoustic features (PAFIND) values obtained from various Asiatic PWS skin conditions by PAMB, where bin interval = 6 and n = 30 for each distribution. (a) Histogram of lesion areas from faces and lip lesions. Green dashed line is Gaussian fit for face lesion, with mean of 38.09 and standard deviation (SD) of 6.92. Red dashed line is Gaussian fit for lip lesion, with mean of 83.09 and SD of 7.51. (b) Histogram of lesion areas from central and lateral face lesion. Green dashed line is Gaussian fit for central face lesion, with mean of 32.83 and SD of 6.67. Red dashed line is Gaussian fit for lateral face lesion, with mean of 48.95 and SD of 7.14. (c) Histogram of lesion areas from different stages. Green dashed line is Gaussian fit for pink lesion, with mean of 19.23 and SD of 7.01. Red dashed line is Gaussian fit for purple lesion, with mean of 47.14 and SD of 7.25. Blue dashed line is Gaussian fit for proliferative lesion, with mean of 73.88 and SD of 7.39. (d) Histogram of lesion areas from different ages groups. Green dashed line is Gaussian fit for 5–15 years old, with mean of 29.1 and SD of 6.77. Red dashed line is Gaussian fit for 16–30 years old, with mean of 50.94 and SD of 7.74. Blue dashed line is Gaussian fit for 31–50 years old, with mean of 75.79 and SD of 7.76. The Gaussian fit can be verified by Shapiro-Wilk (S-W) test.

Download Full Size | PDF

Table 4. The correlated parameters of blood vessels of different diagnostic types of PWS and healthy skin as standard (Ā±SD) by PAMB system^a

View Table | View all tables in this article

4. Discussions and conclusions

The PAM is a promising technique for the noninvasive observation of micrvascular structure and skin disease detection. As light is absorbed by melanin and haemoglobin of skin and converted to heat, the subsequent thermoelastic expansion generates an acoustic wave. The morphological characteristics of the epidermis and dermis in skin can be imaged with a rich optical contrast and high ultrasonic resolution. PAM opens a new chapter in the morphological examination of pigmented and vascular dermatosis as it can improve the accurate diagnoses of skin disease and can be used to quantitatively evaluate the curative effect of treatment in clinical practice and studies [46–56].

We developed a PAMB system that simultaneously integrated high spatiotemporal resolution, deep penetration, and high detection sensitivity for the quasi-histological imaging of PWS. The adjustable confocal opto-sono objective is essential for the imaging performance of PAMB. The best results are achieved by adjusting the depth of focus, which enables the analysis of the fine features visualised using PAMB. Broadband PVDF detection enabled axial and lateral acoustic resolutions of 34 μm and 92 μm, respectively (Supplement 1 Notes S1–S2), which remained approximately constant throughout the entire dermis (1.5 mm deep) and were slightly reduced in deeper layers. For example, the axial resolution was ∼40 μm at 2 mm below the surface (Fig. 1(c)). PAMB captures the natural differences in light absorption in tissues, which allows for the direct and highly sensitive detection of blood vessels (haemoglobin) in human skin, thereby creating an alternative diagnosis method of the pathophysiology.

In addition to demonstrating the imaging performance, our study revealed some critical findings. The PAMB can offer a capacity for imaging different types of cells, visualising epidermis, and measuring and quantifying skin morphology and blood capillary landmarks. These capabilities are not available in other modalities currently used in clinical practice. In addition, vascularisation in the dermis is believed to play a key role in the development of PWS, and a larger number of disease biomarkers have been summarised by PAFIND to serve as a quantitative measure of disease severity.

Pilot comparisons between PAFIND and EI revealed a good correlation; however, EI method can't achieve a fair measurement for different parts of each patient owing to the inability to unify the environmental conditions in the process of photographing. Moreover, it is through the macroscopic comprehensive reflection of the reflected light intensity on the tissue surface, cannot obtain the spatial information of blood vessels structure. Thus some differences between the two indices are to be expected, given that PAFIND considers blood vessels as structural features. By contrast, EI only assessed the colour intensity of PWS. Measurements in patients with PWS also captured changes in the capillary-loop depth and diameter and vascular volume, making it possible to perform an objective clinical evaluation of the severity of PWS or curative effect of treatment. Thus, inter-observer variation is avoided.

Adding the ability to obtain spectral measurements will be an important next step in the development of a portable PAMB. For wide-field fast-scanning imaging, the use of ultra-fast pulsed lasers will be essential for the operation of this PAMB system based on the use of low-pulse-repetition-rate DTL lasers with pulse widths ≥ 4 ns, which significantly affects the imaging speed and is very expensive for clinical use in dermatology. Furthermore, using a single wavelength may be a major limitation – the melanin and blood are differentiated mainly by spatial distribution and structure. A two-wavelength approach would presumably work more effectively to differentiate melanin from blood vessels. Additionally, to obtain a greater depth, a longer wavelength irradiation source may be feasible, indicating that multi-wavelength applications are also potentially attainable. This feature could also enhance the clinical applications of the system by allowing for measurements of oxyhaemoglobin saturation.

In summary, the research could solve some important scientific problems. First of all, the relationship among the characteristics of photoacoustic imaging, the pathological changes and the color types of clinical manifestations could be defined by obtaining the photoacoustic morphological features of PWS (such as epidermal thickness, melanin content in epidermis, dermal thickness, and diameter, density, and depth of the abnormal vessels in dermis, etc.). Secondly, the image criteria could be established for objective classification and quantitative evaluation of PWS by analyzing the differences of vascular parameters among different color types. Besides, the relationship between pathophysiological mechanisms and curative effect could be made clear through the self-controlled study of PWS on photodynamic therapy, providing objective basis for the selection of treatment programs, optimization of treatment parameters and prognostic evaluation of PWS. Thus, the use of PAMB has been successfully translated from the bench to the bedside. This novel medical imaging system has the potential to be used to understand the processes involved in the onset and progression of skin disease, and to monitor treatment options simultaneously. PAMB is therefore conducive to quantitatively evaluating skin disease severity, monitoring progression, estimating the prognosis and effectively guiding therapy.