Results

The results of this study are presented in three main parts: (i) the evaluation of the fabricated vascular finger phantom and its ability to reproduce physiologically relevant PPG signals across three skin pigmentation levels, (ii) analysis of photon-transport behavior obtained from Monte Carlo simulations, and (iii) the correlation between experimental and simulated results to validate the influence of skin pigmentation on optical PPG performance.

The in vitro experiments were conducted to verify that the advanced finger phantom could generate PPG signals in both reflectance and transmittance configurations under pulsatile flow conditions. The Monte Carlo simulations were then used to interpret the optical behavior observed experimentally, specifically, how the different skin layers influence photon transport, penetration depth, absorption, and detected intensity.

In Vitro Experiment

The fabricated finger phantom enabled continuous fluid perfusion through the three-layer channel vascular network without any leak or occlusion. Pulsatile flow at 60 bpm was successfully generated using a programmable pump, producing pressure variations that resulted in detectable optical modulations at the fingertip. Figure 7 shows the experimental setup used for both reflectance and transmittance configurations, including the cardiovascular flow loop, the three interchangeable skin tone layers representing pale, medium, and dark skin pigmentation, and the finger phantom connected to the circuit.

Fig. 7: In vitro experimental setup. (a) Cardiovascular flow loop used to perfuse the vascular finger phantom under physiologically representative pulsatile flow conditions, incorporating the pulsatile pump, ascending aorta, brachial artery, radial artery, and returning venous pathway. (b) The three interchangeable skin tone layers represents pale, medium, and dark pigmentation. (c) The vascular finger phantom with a skin tone layer.

PPG signals were successfully recorded for all three skin tones at 530, 665, and 940 nm in both acquisition modes. The waveforms exhibited characteristic physiological PPG signals. Figure 8 presents a 10-s window of the recorded PPG signals for pale, medium, and dark skin tones in reflectance (left) and transmittance mode (right), illustrating differences in signal morphology and confirming the finger phantom’s ability to generate pulsatile signals in both measurement modes.

Fig. 8: Comparison of reflectance (left) and transmittance (right) PPG signals from the vascular finger phantom across three skin tone conditions: pale (yellow), medium (turquoise), and dark (purple). Each plot shows a 10-s window of the PPG signal for the red (top), infrared (middle), and green (bottom).

Quantitative assessment was performed using the peak-to-peak AC amplitude, SNR, and AC/DC ratio, which together characterize PPG signal quality and sensitivity to skin pigmentation. Table 3 summarizes the median and interquartile range (IQR) of these metrics across wavelengths and skin tones.

In reflectance mode, clear PPG waveforms were recorded across all wavelengths and skin tones, confirming the finger phantom enabled reliable optical detection of pulsatile blood-volume modulation. As illustrated in Figs. 9(a)–9(c), infrared produced the largest peak-to-peak amplitudes and the most stable signals, followed by red and then green wavelengths. Quantitatively, infrared reflectance amplitudes decreased from a median of 8.57 V for pale skin to 5.59 V for dark skin, while red amplitudes showed a similar attenuation from 6.89 to 3.09 V (Table 3). Green reflectance signals were substantially weaker and exhibited the strongest sensitivity to skin pigmentation, decreasing from 2.01 V in pale skin to 0.33 V in dark skin.

In transmittance mode, the observed trends differed [Figs. 9(d)–9(f)]. As expected, the green wavelength produced negligible signals across all skin tones, reflecting its shallow penetration depth and predominant absorption in superficial layers, preventing light from reaching the deeper vascular regions. In contrast, red and infrared wavelengths, which were the target wavelengths for transmittance measurements, produced detectable PPG signals. Their amplitudes were lower than those in reflectance mode, reflecting the longer optical path through the phantom. Infrared exhibited the highest transmittance signal amplitude, with median amplitudes decreasing from 2.80 V for pale skin to 1.95 V for dark skin, while red amplitudes decreased from 0.65 to 0.27 V (Table 3). Increasing skin pigmentation resulted in a consistent reduction in transmittance signal amplitude, reflecting higher absorption within the dark pigmented skin layers.

 | Pale skin | Medium skin | Dark skin
Skin layer |  | Median | IQR | Median | IQR | Median | IQR
AC Amplitude (V) (n=298)
Reflectance | Red | 6.891 | 0.081 | 5.687 | 0.089 | 3.094 | 0.052
IR | 8.567 | 0.116 | 7.712 | 0.101 | 5.588 | 0.105
Green | 2.005 | 0.037 | 0.763 | 0.031 | 0.330 | 0.022
Transmittance | Red | 0.647 | 0.043 | 0.428 | 0.026 | 0.271 | 0.026
IR | 2.795 | 0.045 | 2.495 | 0.034 | 1.951 | 0.036
Green | 0.074 | 0.037 | 0.056 | 0.023 | 0.063 | 0.034
SNR (dB) (n=60)
Reflectance | Red | 37.410 | 0.540 | 37.818 | 0.592 | 35.217 | 1.039
IR | 38.081 | 0.417 | 39.114 | 0.367 | 38.267 | 0.585
Green | 28.708 | 0.599 | 22.083 | 1.009 | 15.437 | 1.593
Transmittance | Red | 17.123 | 0.711 | 16.581 | 1.136 | 11.980 | 1.193
IR | 32.570 | 0.626 | 32.387 | 1.448 | 29.941 | 0.935
Green | −1.799 | 1.704 | −2.948 | 1.696 | −2.960 | 1.625
AC/DC (%) (n=297)
Reflectance | Red | 5.280 | 0.062 | 4.673 | 0.073 | 3.706 | 0.062
IR | 5.887 | 0.081 | 4.920 | 0.064 | 4.252 | 0.080
Green | 6.160 | 0.113 | 4.962 | 0.203 | 3.542 | 0.234
Transmittance | Red | 5.358 | 0.350 | 5.415 | 0.325 | 4.181 | 0.399
IR | 5.648 | 0.090 | 5.956 | 0.082 | 5.078 | 0.095
Green | 2.575 | 1.294 | 2.070 | 0.856 | 2.361 | 1.252

Signal quality trends were further supported by the SNR and AC/DC ratio metrics (Fig. 9 and Table 3). In reflectance mode infrared achieved the highest SNR values across all skin tones (∼38 to 39 dB), followed by red, while green exhibited a marked decrease in SNR with increasing pigmentation, with ∼28 dB in pale skin and falling to ∼15 dB in dark skin. In transmittance mode, SNR values were lower overall due to the increased optical path length, with infrared remaining the most robust wavelength (∼33 dB in pale skin, decreasing to ∼30 dB in dark skin), followed by red (∼17 dB in pale skin, decreasing to ∼12 dB in dark skin). Green transmittance signals exhibited no pulsatile content across all skin tones.

The AC/DC ratio exhibited similar behavior for the wavelengths and skin pigmentation effects. In reflectance mode, AC/DC ratios decreased progressively with increasing skin pigmentation for all wavelengths, with the largest reductions observed for green light. In transmittance mode, infrared and red wavelengths maintained comparable AC/DC ratios across skin tones, while green showed increased variability, consistent with its limited penetration depth.

Fig. 9: Boxplot comparison for PPG signal quality metrics across skin tones and wavelengths. Peak-to-peak AC amplitude, signal-to-noise ratio (SNR), and AC/DC ratios are shown for pale (yellow), medium (turquoise), and dark (purple) skin layers under reflectance (a)–(c) and transmittance (d)–(f) measurement configurations. Boxes indicate median and interquartile ranges, with red asterisks that denote outliers.

Collectively, these results demonstrate that the improved perfusion phantom produces physiologically realistic PPG waveforms across multiple skin tones and wavelengths for reflectance and transmittance mode.

Monte Carlo Simulation Results

Simulations incorporating pale, medium, and dark skin tones were performed using the experimentally derived optical properties of the fabricated phantom materials (Table 2). By varying only the optical properties of the skin layer while keeping tissue geometry and adipose and blood optical properties constant, the simulations isolate the effect of skin pigmentation.

Simulations were conducted at 530, 665, and 940 nm under identical geometry and illumination conditions, enabling direct comparison of photon transport and energy deposition metrics across wavelengths and skin tones.

Photon transport and energy deposition behavior

Figure 10 shows the simulated photon fluence distributions for pale, medium, and dark skin tones at 530, 665, and 940 nm in reflectance and transmittance modes. The tissue depth (z-axis) and lateral distance (x-axis) correspond to the simulated tissue region, with dashed lines indicating the interfaces between layers. In reflectance, photon energy was concentrated near illumination site, with shallow interaction at 530 nm, extending slightly deeper at 665 nm, and reaching the middle blood layer at 940 nm. In transmittance, green light was fully attenuated within superficial layers, while limited transmission was observed at 665 nm, and pronounced propagation through the full tissue depth occurred at 940 nm.

Fig. 10: Simulated photon fluence distributions for pale, medium, and dark skin tones at 530, 655, and 940 nm in reflectance (left) and transmittance (right) modes.

Across all wavelengths, increasing skin pigmentation reduced the overall fluence magnitude due to higher skin absorption. Although these differences are visually subtle, they are consistent with the expected attenuation behavior and support the wavelength-dependent transition from shallow to deep optical interaction within the finger phantom model.

To quantify this behavior, Fig. 11 summarizes the distribution of absorbed photon energy across tissue layers and the resulting skin contribution index (SCI). At 530 nm, absorption was dominated by the skin and upper blood layers, together accounting for more than 70% of total absorbed energy. At 665 nm, absorption extended further into the adipose and middle blood layer regions, while at 940 nm, absorption became more evenly distributed throughout the tissue depth, reflecting enhanced photon penetration.

The SCI decreased with increasing wavelength and increased with skin pigmentation, confirming that darker skin retains a larger fraction of absorbed energy within the skin layers across all wavelengths. Dark skin exhibited the highest SCI values, reaching nearly 40% at 530 nm compared with lower contributions for medium and pale skin tones, 37% and 32%, respectively.

Fig. 11: Wavelength and pigmentation-dependent absorption behavior obtained from Monte Carlo simulations. The bar plots show the layer-specific absorbed energy fractions for pale, medium, and dark skin tones at 530, 665, and 940 nm. The line plot (right) quantifies the Skin Contribution Index (SCI), defined as the percentage of total absorbed energy within the pale (blue), medium (yellow), and dark (red) skin tones.

Energy balance and photon transport metrics

Figure 12 summarizes the simulated photon energy balance for pale, medium, and dark skin tones across all wavelengths. Absorption dominated the photon energy balance, decreasing progressively as wavelength increased, from ∼56% at 530 nm to 32% at 940 nm for the dark skin. Reflectance increased slightly with wavelength, indicating deeper photon penetration and reduced superficial attenuation. Transmittance remained negligible at 530 and 665 nm and increased slightly at 940 nm.

Fig. 12: Monte Carlo-derived photon transport metrics across wavelengths (530, 665, and 940 nm) for pale (blue), medium (yellow), and dark (red) skin tones: (a) absorbed energy fraction, (b) diffuse reflectance, (c) transmittance, and (d) fluence-based 50% cumulative absorption depth (Z50).

To characterize photon penetration, fluence-based photon penetration depth was quantified using the 50% cumulative absorption depth. This metric represents the depth at which most photon energy is deposited within the tissue layers. The results show that photon penetration increased with wavelength. At 530 nm, half of the total absorbed energy was confined within the upper 1 mm, extending to around 1.3 mm at 665 nm and 1.5 mm at 940 nm. Across all wavelengths, darker pigmentation produced slightly shallower penetration depths due to the increased absorption, especially in the visible range.

Correlation between Experimental and Simulated Data

Strong agreement was observed between Monte Carlo-simulated optical trends and experimentally measured PPG metrics across wavelengths and skin tones. In reflectance mode, AC/DC ratios decreased consistently with increasing skin pigmentation at all wavelengths, yielding strong negative correlations (|r|≥0.94). In transmittance mode, negative correlations were observed at 665 and 940 nm (|r|≈0.8), while no reliable PPG signal was detected at 530 nm due to the shallow penetration depth of green light, and this wavelength was excluded from quantitative interpretation [Fig. 13(a)].

Simulated detected intensity showed strong positive agreement with experimentally measured PPG peak-to-peak amplitude. In reflectance mode, correlations were consistently high across all wavelengths (r≥0.98), indicating that pigmentation-driven reductions in simulated diffuse reflectance translated directly into reduced signal amplitude. In transmittance mode, similarly strong agreement was observed at 665 and 940 nm (r≈0.98), whereas the 530 nm transmittance mode was excluded due to insufficient signal presence [Fig. 13(b)].

Fig. 13: Correlation between Monte Carlo simulations and in vitro PPG measurements. (a) Relationship between Skin Contribution Index (SCI) and AC/DC ratio for reflectance and transmittance modes across wavelengths. (b) Relationship between Monte Carlo simulated detected intensity and experimentally measured PPG peak-to-peak amplitude. Data points represent pale, medium, and dark skin tone layers of the finger phantom, solid lines indicate linear fits, and Pearson’s r-values are reported. The 530 nm transmittance results are shown for completeness but excluded from quantitative interpretation due to insufficient signal, given the shallow penetration depth of this wavelength.

Overall, the correlation analysis demonstrates that Monte Carlo simulated changes in photon transport with increasing skin pigmentation are closely reflected in both PPG amplitude and AC/DC ratio, with the strongest and most consistent agreement observed in reflectance mode. Although the correlation analysis is descriptive given the three data points representing the three skin tone conditions, the consistent direction and magnitude of correlations across wavelengths and measurement modes suggest that the observed relationships between simulated photon transport and experimental PPG metrics are robust and representative of the underlying optical behavior.