Perceptibility and acceptability thresholds of simulated facial skin color differences

This study investigates the perceptibility and acceptability thresholds (PT, AT) of observers for L*, a* and b* facial skin differences, very important for the fabrication of a maxillofacial prosthesis.

One image (10×13.3cm/300dpi), simulating standard facial skin color, and 27 smaller images (4.9×6.0cm/300dpi), simulating shades with different RGB values from the standard were made, using Adobe-Photoshop CS3. The RGB values, were based on 9 stepped increases in L*, a* and b*. Images were shown in the middle of a computer screen with the standard image in the background. 90 persons with normal color vision and optical acuity were selected and instructed to evaluate 81 (27×3) randomly shown images for a perceptible or an acceptable (48h later) mismatch, between central and background image. Using best fitting regression curves 50 % PT and 50 % AT for all parameters were estimated along with their 95 %CI.

Observers' reliability found high (0.807-0.945) while the cubic and exponential were the best fitted curves to the data. Overall 50 % ATs were smaller than 50 % PTs (p<0.01) and overall ΔL* parameter showed the lowest thresholds (0.119PT/0.864AT), significantly smaller (p<0.05) than Δa* (0.993PT/2.037AT), Δb* (1.147PT/2.517AT), and ΔEab*(1.497PT/1.808AT). No differences were found between sex or age groups in 50 % ATs for Δa* and Δb* parameters, but 50 % PT for Δa* was smaller in females and in older people with the opposite for Δb*.

Changes in L* were more perceptible and less acceptable than changes in a* and b*. For this reason ΔL* parameter needs attention equally to ΔE*ab when fabricating a facial prosthesis.