Comparison of body fat estimates using 3D digital laser scans, direct manual anthropometry, and DXA in men

Abstract

The purpose of this study was to assess the feasibility of utilizing three dimensional whole body laser surface scanning (3DS) to obtain specific anthropometric measurements to estimate percent body fat (BF). Percent BF estimates from 37 male volunteers, of age 18-62 yr, were determined by inputting manual anthropometric (MA) and 3DS anthropometric measurements into the current Army BF prediction equation for males. The results were compared with each other and to BF values from Dual Energy X-ray Absorptiometry (DXA), employed as a reference method. Mean percent BF estimates (+/-SD) derived from MA, 3DS and from DXA were 18.4(+/-3.8), 18.8(+/-3.9), and 18.9(+/-4.7), respectively. Analysis of Variance tests revealed no statistical difference between the mean values. Correlation analysis comparing MA and 3DS derived percent BF estimates to each other and to those measured by DXA revealed moderate to strong Pearson correlation coefficients (r), small to moderate standard errors of the estimate (SEE), and were statistically significant (p < 0.05). Correlation coefficients and SEE results for this sample were: (1) DXA vs 3DS; r = 0.74, SEE = 3.2, (2) MA vs DXA; r = 0.82, SEE = 2.8, and (3) MA vs 3DS; r = 0.96, SEE = 1.0. Lin's concordance analysis, including Bland-Altman limits of agreement (LOA), revealed statistically significant measurement agreement among the three measurement modalities (p < 0.05). The application of 3DS scanning to estimate percent BF from commonly used anthropometric measurements are in close agreement with BF estimates derived from analogous MA measurements and from DXA scanning.