2.1. Design Requirements for Medical Lighting

Medical lighting must provide sufficient and uniformly distributed illuminance. Here, the average illuminance is defined as the arithmetic mean over the observation plane, and uniformity is defined as the ratio of minimum to average illuminance on that plane. EN 12464-1 and KS A 3011 specify at least 200 lx with a uniformity of 0.6 or higher for hospital wards^{[11, 12]}. The proposed system is intended as ceiling-mounted lighting for general wards and inpatient observation rather than for high-illuminance treatment or emergency use; its design parameters were therefore set to meet general-ward requirements across the entire tunable CCT range, while higher levels may be required in environments such as emergency or treatment rooms. To ensure accurate perception of skin and mucous membrane color changes, an Ra of 90 or higher was adopted as the baseline. Ra, Rf, and Rf_skin were used to assess general color rendering, color fidelity, and skin-color reproduction, respectively, with Rf and Rf_skin evaluated by the ANSI/IES TM-30 method^[13]. In addition, cyanosis observation performance was assessed using the CIEDE2000-based CCDI, which quantifies the perceptual difference between normal and cyanotic skin rather than applying the fixed COI threshold of the AS/NZS standard.

2.2. Spectral and Color Characteristics of LED Light Sources

The Luminus MP-3030-21C2 LED package was employed in this research. The main specifications of the LED package, including package size, nominal CCT, Color Rendering Index (CRI), and relevant optical characteristics, are summarized in Table 1, which is based on the manufacturer’s datasheet^[14].

To implement the dual-channel configuration, 3000 K and 5000 K LEDs were utilized as they exhibit distinct spectral characteristics. The selected LED package provides high color rendering performance (CRI ≥ 90) and broad spectral characteristics, which are important for perceiving subtle color changes in the skin and mucous membrane.

Fig. 1. Chromaticity coordinates and spectral power distributions of 3000 K and 5000 K LED light sources

Fig. 1 illustrates the CIE color coordinates and SPDs of the two LEDs. The 3000 K LED exhibits high radiant intensity near 630nm, emphasizing the red spectral region. In contrast, the 5000 K LED shows pronounced spectral components around 450nm and 490nm, highlighting the blue–cyan spectral range.

Specifically, in the 450–500nm range, the radiant intensity of the 5000 K LED is approximately 2 to 3 times higher than that of the 3000 K LED, which contributes to enhancing the reflection contrast between oxygenated and deoxygenated blood. Conversely, the 3000 K LED shows a relatively high radiant intensity in the red spectral region, making it advantageous for reproducing the red components of the skin and mucous membrane.

Table 2 summarizes the color quality metrics for the two LEDs. The measured CCTs were 2991 K and 4827 K, respectively, which are close to the target CCTs. Furthermore, the Ra of both light sources satisfies the standards for medical lighting, and the Rf and Rf_skin also exhibit similar levels.

As such, the two light sources exhibit complementary spectral characteristics in the blue–cyan and red regions. Their combined operation provides a foundation for supporting both cyanosis contrast perception and skin color reproduction across the tunable CCT range.

2.3. Dual-Channel LED Panel Design

Fig. 2 illustrates the dual-channel LED panel and the simulation configuration. The proposed light source is designed with a panel structure of 594mm × 594mm × 31mm and consists of a total of 140 LEDs.

Fig. 2. Configuration of the dual-channel LED panel and simulation setup, including panel dimensions, LED array (3000 K and 5000 K), and observation plane geometry

The optical and electrical characteristics of the dual-channel LED panel, based on the MP-3030-21C2 LED package, are summarized in Table 3. The LED package specifications were obtained from the manufacturer’s datasheet^[14], and the panel-level driving conditions were determined based on the LUT-based current control used in the simulation.

The 3000 K and 5000 K LEDs are arranged in an alternating pattern to ensure uniform distribution across the panel surface. This layout is intended to achieve illuminance uniformity through the spatial mixing of the two light sources during tunable CCT operation. The same configuration is applied as the simulation conditions for the subsequent optical performance evaluation, where the observation plane is defined as a 2.12m × 2.12m area located 2m below the panel.

2.4. Lookup Table (LUT)-Based CCT Control and Current Configuration

Table 4 presents the LUT summarizing the driving currents for the 3000 K and 5000 K channels according to the target CCT, along with the resulting simulated CCT and the corresponding error (ΔCCT).

As the CCT increases, the driving current for the 3000 K channel decreases from 1286mA to 0mA, while the current for the 5000 K channel increases from 0mA to 1054mA. In the intermediate CCT range, both channels are driven simultaneously; for example, at a target CCT of 4000 K, the current is distributed as 469mA and 669mA for each channel, respectively. The simulated CCT exhibits an error ranging from -185 K to +90 K relative to the target values, which corresponds to a relative error of 0.26– 3.7%. This LUT-based current control was implemented to compensate for the nonlinear spectral mixing characteristics associated with CCT variations and to achieve stable operation across a wide CCT range.

3.1. Illuminance Distribution and Uniformity Characteristics

Optical performance evaluation was performed using LightTools ray-tracing analysis based on the simulation conditions presented in Fig. 2. The luminous intensity characteristics and SPDs of the LEDs were based on datasheet values, and the light output for each CCT condition was calculated by reflecting the LUT-based current settings from Table 4.

Fig. 3 shows the ray-tracing results, luminous intensity distribution, and illuminance distribution at the 4000 K condition. The luminous intensity distribution exhibits a diffusive characteristic of 113°; consequently, the emitted rays are distributed across the entire observation plane. The illuminance distribution on the observation plane shows high values at the center and decreases towards the periphery.

Fig. 3. Ray-tracing visualization, angular light distribution, and spatial illuminance distribution of the dual-channel LED module at 4000 K on the observation plane

Fig. 4 illustrates the variations in average illuminance and uniformity across the entire target CCT range (3000–5000 K) based on the LUT. The average illuminance is maintained within the range of 215.2–224.1lx, exhibiting a variation of 4.06% throughout the CCT range. Uniformity is maintained between 0.754 and 0.760, showing a minimal variation of 0.79%. This indicates that the light output and spatial distribution are stably preserved as the currents of the two channels are complementarily adjusted in response to CCT changes.

Fig. 4. Average illuminance and uniformity as a function of correlated color temperature (CCT) under LUT-based current control

In addition, the optical performance values at representative CCT conditions (3000 K, 4000 K, and 5000 K) are summarized in Table 5. The average illuminance was calculated at 215lx, 223lx, and 215lx, respectively, satisfying the hospital lighting requirement of 200lx or higher. The uniformity ranged from 0.754 to 0.760, exceeding the 0.6 criterion specified by EN 12464-1 and KS A 3011.

3.2. Evaluation of Color Quality and Cyanosis Observation Performance

Fig. 5 shows the variations in the color quality metrics and the CCDI across the tunable CCT range.

Fig. 5. Variation of color quality indices (Ra, Rf, Rf_skin) and the cyanosis color-difference index (CCDI) as a function of correlated color temperature (CCT) under LUT-based current control

Rather than computing the AS/NZS COI directly, a CIEDE2000-based CCDI was designated to quantify how perceptibly cyanotic skin differs from normal skin under each test SPD. The spectral reflectances of normal (oxygenated) and cyanotic (deoxygenated) skin were generated with a single-layer Kubelka–Munk model^{[3, 4]} over 400–700nm at 5nm intervals. The absorption coefficient combined hemoglobin (oxygenated and deoxygenated molar extinction from Prahl^[15]; blood volume fraction 8%, oxygen saturation 98% for normal and 70% for cyanotic skin), melanin (3%), and a baseline tissue term following Jacques^[3], with a skin reduced-scattering power law^[3]; the diffuse reflectance followed the Kubelka–Munk relation. For each SPD, both skin conditions were rendered under the same illuminant (no external reference), and the CIEDE2000 difference between them was computed^[16]:

where Lab_normal and Lab_cyanotic are the CIELAB coordinates of the modeled normal and cyanotic skin under the test SPD. A larger CCDI indicates a greater perceptual difference, i.e., more readily observable cyanosis; unlike the AS/NZS COI (lower = less distortion), it is a contrast measure for which larger values are favorable, so the 3.3 criterion does not apply.

Under tunable operation, Ra, Rf, and Rf_skin remained stable, with variations of 1.6% (92.3–93.8), 0.81% (89.5–90.2), and 2.16% (91.3–93.3), respectively. The CCDI increased monotonically from 0.86 at 3000 K to 1.54 at 5000 K, indicating that cyanosis is least perceptible at 3000 K and most at 5000 K, consistent with the higher 450–500 nm content of the 5000 K channel. Conversely, skin redness a* (evaluated against each SPD’s own white point) decreased from 15.4 to 12.4, indicating more natural skin-color reproduction at a lower CCT. Since the two move in opposite directions, no single fixed CCT maximizes both; the dual-channel tunable configuration—combining the cyanosis-contrast advantage of the 5000 K channel with the skin-color advantage of the 3000 K channel—lets the operator select the CCT best suited to each clinical situation. Table 6 summarizes the representative results.

To relate the LED sources to the panel-level simulation, the measured color quality indices were compared with the simulated values at representative CCTs (Table 7).

The measured and simulated Ra, Rf, and Rf_skin agreed closely at both 3000 K and 5000 K, indicating that the source characteristics were well reflected in the panellevel simulation and that the proposed tunable system can flexibly select CCT through active SPD control.