Key points are not available for this paper at this time.
PURPOSE: To explore the feasibility of a novel computer color-matching (CCM) system based on the improved back-propagation neural network (BPNN) model by comparing it with the traditional visual method. MATERIAL AND METHODS: Forty-three metal-ceramic specimens were fabricated by proportionally mixing porcelain powders. Thirty-nine specimens were randomly selected to train the BPNN model, while the remaining four specimens were used to test and calibrate the model. A CCM system based on the improved BPNN model was constructed using MATLAB software. A comparison of the novel CCM system and the traditional visual method was conducted by evaluating the color reproduction results of 10 maxillary central incisors. Metal-ceramic specimens were fabricated using two color reproduction approaches. Color distributions (L*, a*, and b*) of the target teeth and of the corresponding metal-ceramic specimens were measured using a spectroradiometer. Color differences (ΔE) and color distributions (ΔL*, Δa*, and Δb*) between the teeth and their corresponding specimens were calculated. RESULTS: The average ΔE value of the CCM system was 1.89 ± 0.75, which was lower than that of the visual approach (3.54 ± 1.11, p 0.05). CONCLUSION: The novel CCM system produced greater accuracy in color reproduction within the given color space than the traditional visual approach.
Wei et al. (Wed,) studied this question.