Method: Six linear measurements [total facial height (N-Gn), face width (Zy-Zy), lower face width (Go-Go), nasal width (Al-Al), nose height (N-Sn), and lower lip height (Sto-Gn)], and six facial proportions (N-Gn/Zy-Zy, Al-Al/N-Sn, Sto-Gn/Go-Go, Go-Go/Zy-Zy, Go-Go/N-Gn, Sto-Gn/N-Gn) were obtained from facial moulages of orthodontic patients (n=37) by each of the three techniques. The Bland-Altman method was used to test the equivalence between these techniques. Linear measurements and facial proportions were considered equivalent if the differences did not exceed 1mm and 5%, respectively (p<0.05).
Result: The results showed that 3D and direct anthropometry were equivalent with differences smaller than 0.52 mm and 3% (p<0.0001) in all measurements except in nasal width which showed a mean difference of 1mm. In contrast, the 2D technique showed underestimation greater than 2mm in linear measurements when compared to 3D and direct anthropometry; however, most of the facial proportions were equivalent with the two other methods (p<0.0001) except for Al-Al/N-Sn and Sto-Gn/Go-Go which showed differences between 6% and 9%.
Conclusion: The accuracy of 3D measurements of facial soft-tissue structures is comparable to direct caliper-based anthropometry. On the other hand, conventional 2D facial assessment underestimates linear measurements and should be limited to qualitative facial analysis.
Keywords: Digital image analysis, Facial Anthropometry and Orthodontics