2.1.

Type 1 Diabetes (T1D) Model

Adult male Balb/c mice (8 weeks old, $23\pm 2\mathrm{g}$) were intraperitoneally administered with streptozotocin (STZ, V900890, Sigma, St. Louis, MO) at a dosage of $200\mathrm{mg}/\mathrm{kg}$ at 4 h after fasting. To allocate sodium citrate buffer, 2.1 g citric acid monohydrate (C1909, Sigma, St. Louis, MO) and 2.94 g sodium citrate (S4641, Sigma, St. Louis, MO) were dissolved in 100 ml distilled water, respectively, and then mixed in proportions of 1:1.32, whose PH was 4.2 to 4.5. Each time before injection, the STZ was dissolved in sodium citrate buffer at 1% (w/v). Intraperitoneal injections should be finished within 15 min after drug preparation. The mice were injected for five consecutive days. After multiple STZ injection, the blood glucose was measured using a glucometer (580, Yuyue Medical Equipment & Supply Co., Ltd., China). Mice with fed blood glucose $\ge 16.7\mathrm{mM}$ ($300\mathrm{mg}/\mathrm{L}$) were considered diabetic and selected for imaging. The protocol refers to relevant literature.^22,23 Additionally, the mice were divided into five groups (Normal; DM-1 week; DM-2 weeks; DM-3 week; DM-4 week). There were five mice in each group for PSOCT and SEM imaging. For Masson’s trichrome staining experiments, five mice were used in each group. All experimental procedures were carried out in accordance with the Guangdong Provincial Experimental Animal Management Ordinance and the Guangdong Medical University’s guidelines, which had been approved by the Central People’s Hospital of Zhanjiang’s Institutional Animal Ethics Committee.

2.2.

PS-OCT System

We used a commercial PS-OCT imaging system (TEL221PS, Thorlabs Inc., Germany), with a scanning lens (OCT-LK3, 5×), to monitor how the skin structure changes as DM progresses. The central wavelength is 1300 nm, with a bandwidth of 170 nm. The imaging depth and axial resolution in air/water are $3.5\mathrm{mm}/2.6\mathrm{mm}$ and $5.5\mathrm{mm}/4.2\mathrm{mm}$, respectively. This PS-OCT utilizes an incident beam with known polarization and a dual-detector design to incorporate polarization information in 2D cross-sectional and 3D volumetric images of a sample. It controls the polarization incident on the sample and reflects it in the reference arm of the interferometer by using two carefully aligned wave plates. The preserved polarization information is then measured using two detectors. The schematic of this PS-OCT system is shown in Fig. 1.

Fig. 1

Schematic of PS-OCT imaging system.

2.3.

Evaluation of Polarization Parameters

We utilized the 3D polarization-sensitive mode in the PS-OCT system to capture polarization images and collected intensity and polarization signals from two sensors. The Stokes vectors ($I$, $Q$, $U$, and $V$) can be determined from the data provided by the two orthogonal polarization states of the interference light for each point. The retardation, optic axis, and degree of polarization uniformity (DOPU) are three advanced polarization characteristics that can be calculated using the Stokes vector to describe the polarization state of light. The formula below can be used to evaluate the polarization parameters

The intensity of detector-0 and detector-1 is represented by $I_{0/1}$. $Q$ stands for the percentage of either horizontally ($Q=-1$) or vertically ($Q=+1$) polarized light. $U$ is proportion of light linearly polarized at 45 deg ($U=-1$) or linearly polarized at 135 deg ($U=+1$). $V$ is proportion of left-circularly ($V=-1$) or right-circularly ($V=+1$) polarized light. The algorithm specifics are described previously.²⁴ ThorImageOCT (version 5.4) is utilized for acquiring and processing polarization images.

Additionally, intrinsic birefringence mainly occurs in regularly ordered fibrillar tissues²⁵ The local birefringence ($\mathrm{\Delta }n$) can be determined from the cumulative phase retardation. It is calculated according to the following equation:²⁶

The $\lambda$ is center wavelength of the light used. The $d{\delta }_{\mathrm{cum}}(z)/dz$ is the local slope of cumulative phase retardation as a function over $z$ position.

2.4.

SEM for Obtaining the Microstructure of Skin with Development of T1D

For the normal and T1D mice at different stages, $1{\mathrm{cm}}^2$ skin samples were collected after using a hair removal cream (3221469, Veet, Reckitt Benckiser Group Plc (RB), France), and PBS ($>4$ times) carefully washed the residual blood. Then, these samples were fixed with 2.5% glutaraldehyde (G5882, Sigma, St. Louis, MO) overnight at 4°C. The samples were rinsed twice in PBS and then fixed for an hour in 1% osmium acid that had been pre-cooled to 4°C. The samples were dehydrated, rinsed twice with PBS, and then treated with dry heat. Finally, the samples were processed by conduction, and a scanning electron microscope (SU8100, Hitachi, Ltd., Japan) was used to capture images of the skin’s microstructure. To determine the structure of collagen fibers in the skin dermis, it was necessary to scan the cross-section of the skin sample at a magnification of 500. According to the cross-sectional image, the microscopic structure of the collagen fibers was scanned. About $\sim 250\mu \mathrm{m}$ below the epidermis (magnification = 15k/40k). Additionally, we also evaluated the orientation of the skin collagen fibers using the ImageJ/FIJI directionality plugin.

2.5.

Masson’s Trichrome Staining

The collagen in the skin of normal mice was observed through pathological sections at different time points after the application of hair removal cream. The dorsal skin tissue was cut into $\sim 0.5\mathrm{cm}\times 0.5\mathrm{cm}$ pieces and then promptly washed with cool saline to remove blood. The samples were then fixed in 4% neutral paraformaldehyde, dehydrated, and embedded in paraffin. After cutting into $4\text{-}\mu \mathrm{m}$ sections, they were stained with Masson’s trichrome.

3.1.

Effect of Structural Modifications on Polarization State in Multi-Layered Model

The birefringence of standard adhesive tapes was evaluated in a recent study.²⁷ Here, a multi-layered tape made of polyethylene and adhesive (each layer $\sim 0.05\mathrm{mm}$ thick) was used to simulate layered structure of biological tissues. The laminated structure will deform under compression, and we used this multi-layered phantom to simulate the impact of structural modifications on polarization [Fig. 2(a)]. The polarization images were captured before and after deformation. Figure 2(b) shows the typical cross-sectional images. The average residual (the distance from the data to the fitted curve) was calculated to assess the fluctuation amplitude of the corresponding polarization parameters (Table S1 in the Supplementary Material). In the undeformed phantom, there was a consistent change in retardation, optic axis and DOPU. The deformed phantom exhibited significant changes in retardation, optic axis and DOPU. Additionally, we also assessed the local birefringence. The profiles of changes in retardation, optic axis, and local birefringence at different depths indicated that the residual decreased following the deformation. It indicated that the fluctuation amplitude decreased. The Stokes parameters in pixel-neighborhood were homogeneous before deformation, and the phantom refractive index was also uniform, as the DOPU was reasonably close to 1. The profiles of DOPU change at various depths also demonstrated that the magnitude of fluctuation decreased after the deformation. The compression-induced deformation could lead to changes in the original compact and ordered layer structure of the tape, and there were even gaps between some layers. The disorganized distribution of layers and presence of gaps affected light propagation, leading to changes in polarization characteristics. These results suggest that PS-OCT imaging can effectively assess changes in structure of layered tissues.

Fig. 2

PS-OCT imaging of the multi-layered tape model. (a) Schematic diagram of the multilayer tape model before and after deformation. (b) The intensity, retardation, optic axis, and DOPU images of multi-layered adhesive tape model before/after compression (scale bar is $250\mu \mathrm{m}$). The corresponding profiles of the polarization parameters (retardation, optic axis, DOPU, and local birefringence) of the white rectangle in intensity images along the depth direction. The dashed line was the fitted curve for the corresponding data, and the average residual is shown in Table S1 in the Supplementary Material.

3.2.

Changes of Skin Polarization State with Development of DM

Here, we established the STZ-induced T1D model and used PS-OCT to monitor polarization images of the mouse dorsal skin at different stages of DM [Fig. 3(a)]. Figure 3(b) shows the sectional views of intensity, retardation, optic axis and DOPU. The OCT intensity images revealed the structure of skin, including the epidermis, dermis, and hypodermis, with areas of low intensity that likely represent vessels or fat. For the normal mouse, the regular “layer-like structure” was observed in the retardation and optic axis images of skin. The DOPU of epidermis and dermis was higher than that of the hypodermis. According to polarization images, the skin’s polarization characteristics did not change significantly after the diabetic mice were kept in a state of hyperglycemia for 2 weeks. However, for mice in DM-3 week and DM-4 weeks, there were changes in skin retardation and optic axis images, and the “like-layered structures” observed in images of normal skin became disordered.

Fig. 3

Monitoring the dorsal skin of DM mice by PS-OCT. (a) Graphic illustration of the experiment workflow. (b) The intensity, retardation, optic axis, and DOPU images of skin at different stages of DM. (E, epidermis; D, dermis; HD, hypodermis; scale bar is $250\mu \mathrm{m}$.)

Furthermore, based on polarization images collected by PS-OCT, we analyzed profiles of skin polarization characteristics along the depth, as shown in Fig. 4(a). We also assessed the local birefringence based on cumulative phase retardation with the depth. Moreover, the number of peaks and range of values were used to further quantify changes in polarization parameter curves within the skin depth range (depth: 100 to $500\mu \mathrm{m}$) [Fig. 4(b)]. For normal mice, there were multiple peaks in the retardation and optical axis curve, which was similar to the DM-1 week and DM-2 week groups. For DM-3 week and DM-4 week, the retardation and optical axis curves changed significantly, with decreases in range values and the number of peaks. It implied that the amplitude of their fluctuations changed. Additionally, compared with the normal group, there was also a significant decrease in the number of peaks of local birefringence curves in the DM-3 week and DM-4 week groups. The reduction in the fluctuations of the curve might be linked to the changes in skin structure induced by diabetes. This was consistent with the model test results above regarding effects of altered layer structure on polarization parameters. However, when compared with the normal mice, there was no significant change in DOPU curve for diabetic mice. These results suggest that non-invasive PS-OCT imaging can be used to evaluate skin structure disorders caused by diabetes.

Fig. 4

Analysis of DM-induced changes in polarization characteristic parameters (retardation, optic axis, DOPU, and local birefringence) based on the PS-OCT images. (a) The curve and the width of shadow are mean and standard deviation, respectively (E, epidermis; D, dermis; HD, hypodermis; $n=5$, mean ± standard deviation.) (b) The number of peaks and range (max-min) of the curve in (a) (Depth: 100 to $500\mu \mathrm{m}$; $n=5$, error bars represent the standard deviation, one-way ANOVA analysis and two-tail $t$-test. Not significant: $p>0.05$; *$p<0.05$; **$p<0.01$; ***$p<0.001$).

3.3.

Assessment of DM Induced Changes in Skin Collagen

The rich collagen fibers are essential for maintaining the structure and function of skin. The alterations in morphology and diversity of collagen fiber orientation can be effectively revealed through birefringence. Figure 4(a) and Table S2 in the Supplementary Material suggested that, in comparison with normal mice, the local birefringence values showed no significant changes in DM-1 week and DM-2 week mice at the imaging depth of 300 to $500\mu \mathrm{m}$. However, these values decreased significantly for DM-3 week and DM-4 week mice. Accordingly, the results above suggest that changes in collagen fiber structure could be the cause of alterations in skin polarization images. Further, SEM was used to assess the microscopic structure of skin at various stages of DM, as illustrated in Fig. 5(a). To assess the disorder of collagen, we also utilized directionality analysis to evaluate the percentage of collagen fibers oriented 0 deg to 180 deg as shown in Fig. 5(b). It was observed that collagen fibers in the normal, DM-1 week and DM-2 week mice exhibited an orderly alignment, arranged neatly in a relatively consistent direction. However, for DM-3 week and DM-4 week, the distribution of collagen fibers gradually became disorganized, orienting in all directions. These results were consistent with the variations in local birefringence. Furthermore, to determine whether hair removal cream would affect skin’s collagen structure, we conducted Masson’s trichrome staining of the normal skin at various time points following application of the hair removal cream (Fig. S1 in the Supplementary Material). The results indicated that a basic hair removal treatment did not have a significant impact on skin structure. Therefore, it indeed indicates that changes in polarization characteristics could effectively reflect diabetes-induced alterations of skin structure.

Fig. 5

The changes in skin collagen structure at different DM stages. (a) SEM images of skin. (b) Typical quantitative assessment of the collagen orientation from 0 deg to 180 deg. (The polar axis represents the proportion of collagen in each direction.)