Structured light imaging mesoscopy: detection of embedded morphological changes in superficial tissues

Abstract

Significance: Current paradigms for the optical characterization of layered tissues involve explicit consideration of an inverse problem which is often ill-posed and whose resolution may retain significant uncertainty. Here, we present an alternative approach, structured light imaging mesoscopy (SLIM), that leverages the inherent sensitivity of raw spatial frequency domain (SFD) reflectance measurements for the detection of embedded subsurface scattering changes in tissue.

Aim: We identify wavelength-spatial frequency ( $\lambda \text{-}{f}_x$ ) combinations that provide optimal sensitivity of SFD reflectance changes originating from scattering changes in an embedded tissue layer. We specifically consider the effects of scattering changes in the superficial dermis which is a key locus of pathology for diverse skin conditions such as cancer, aging, and scleroderma.

Approach: We used Monte Carlo simulations in a four-layer skin model to analyze the SFD reflectance changes resulting from changes in superficial dermal scattering across wavelength ( $\lambda =471$ to 851 nm) and spatial frequency ( $f_x=0$ to 0.5/mm). Within this model, we consider different values for epidermal melanin concentration to simulate variations in skin tone.

Results: Monte Carlo simulations revealed that scattering changes within the superficial dermis produce SFD reflectance changes which are maximized at specific ( $\lambda \text{-}{f}_x$ ) pairs and vary with skin tone. For light skin tones, SFD reflectance changes due to scattering reductions in the superficial dermis are maximized at $\lambda =621\mathrm{nm}$ and spatial frequency $f_x\approx 0.33/\mathrm{mm}$ . By contrast, for darker skin tones, maximal SFD reflectance changes occur at wavelengths in the near-infrared ( $\lambda \ge 811\mathrm{nm}$ ) at a spatial frequency of $f_x\approx 0.25/\mathrm{mm}$ . Interestingly, the change in SFD reflectance produced by such scattering changes is most uniform across all skin tones when using the longest wavelength tested ( $\lambda =851\mathrm{nm}$ ) and a spatial frequency of $f_x\approx 0.22/\mathrm{mm}$ . Taken together, our computational model identifies specific ( $\lambda \text{-}{f}_x$ ) pairs to optimally detect embedded structural alterations in the superficial dermis.

Conclusions: The findings establish the SLIM methodology as a means to detect morphological changes in an embedded subsurface tissue layer by leveraging inherent sensitivities of spatial frequency domain reflectance. This approach promises to enable simplified clinical tracking of subsurface microstructural alterations without the explicit need to consider an inverse problem approach.

Keywords: Monte Carlo simulation; inverse problems; perturbation methods; spatial frequency domain measurement; subsurface morphological change detection.

Fig. 1

Diagram of the simplified four-layer skin tissue model (adapted from Ref. 43).

Fig. 2

SFD reflectance versus wavelength for baseline (solid black lines) and negative scattering perturbation (dashed red lines) at $f_x=0/\mathrm{mm}$ (*) and $f_x=0.2/\mathrm{mm}$ ($▫$) with 2% epidermal melanin concentration.

Fig. 3

SFD reflectance change $\mathrm{\Delta }{R}_d$ versus spatial frequency $f_x$ at individual wavelengths for epidermal melanin concentrations of (a) 2%, (b) 5%, and (c) 10%. Contour plots showing the variation of SFD reflectance, $\mathrm{\Delta }{R}_d$, with $f_x$ and $\lambda$ for epidermal melanin concentrations of (d) 2%, (e) 5%, and (f) 10%.

Fig. 4

Wavelength dependence of (a) maximum reflectance changes, max $\mathrm{\Delta }R$, and (b) corresponding spatial frequency, $f_x$ at max $\mathrm{\Delta }R$ for 2%, 5%, and 10% epidermal melanin concentrations.

Fig. 5

(a) Dimensionless “optimal” spatial frequency $f_x^{*}$ versus $({\mu }_s^{\prime }/{\mu }_a)$. (b) Spectral dependence of the optimum spatial frequency $f_{x,\mathrm{opt}}$.

Fig. 6

Plots of $p_{z_{\max }}(z)$ for (a) and (b) $\lambda =526\mathrm{nm}$ and (c) and (d) $\lambda =851\mathrm{nm}$ before and after a 50% reduction in scattering within the papillary dermis, respectively, at $f_x=0,0.2$, and 0.5/mm. Bar plots providing sampling depth statistics at (e) and (f) $\lambda =526\mathrm{nm}$ and (g) and (h) $\lambda =851\mathrm{nm}$ before and after scattering a 50% reduction in scattering within the papillary dermis, respectively. The median depth ($d_{50}$) is represented by a circled dot, intervals of 25 to 75% ($d_{25}$ to $d_{75}$) by a rectangle, and 10 to 90% ($d_{10}$ to $d_{90}$) by a capped line at the same spatial frequencies as the $P_{z_{\max }}$ distributions. The shading of the plots is color-coded as follows: blue for the epidermis, green for the papillary dermis, yellow for the reticular dermis, and red for the subcutaneous tissue. All plots are for 2% epidermal melanin concentration. Results for 5% and 10% melanin concentrations are shown in Figs. S1 and S2 in the Supplementary Material, respectively.

Fig. 7

Maximum penetration depth distribution difference before and after scattering perturbation at different spatial frequencies for (a) 526 nm and (b) 851 nm at 2% melanin concentration. Plots are color-coded as follows: blue for the epidermis, green for the papillary dermis, and yellow for the reticular dermis. Results for 5% and 10% melanin concentrations are shown in Figs. S3 and S4 in the Supplementary Material, respectively.

Fig. 8

Individual layer contributions to the total reflectance change at spatial frequencies: $f_x=0$, 0.2, and 0.5/mm for $\lambda =526\mathrm{nm}$ with (a) 2% and (b) 10% melanin concentration and $\lambda =851\mathrm{nm}$ with (c) 2% and (d) 10% melanin concentration. Due to negligible contribution, the epidermis was excluded from the plots. The subcutaneous tissue layer’s contribution is also negligible at shorter wavelengths ($\lambda =526\mathrm{nm}$) due to limited light penetration. Results for 5% melanin concentration are shown in Fig. S5 in the Supplementary Material.