Body measuring tape and skinfold calipers have been used extensively in the past to estimate body composition. However, using measuring tape and skinfold calipers is time consuming and the results may have inter-reader bias. 3D optical (3DO) scanners have become a versatile tool for health assessment and can output automated anthropometry. Accurate and precise equations for children’s body composition have been derived using the automated anthropometry on the Fit3D Proscanner. However, these equations have not been tested on the anthropometry produced by other scanner modalities. The objective of this study was to observe if the equations derived from Fit3D would work on other scanners or if we need to derive scanner specific equations.
Participants (n = 177) were recruited for the Shape Up! Kids study. Each participant received one DXA scan and two optical scans on the Styku and Size Stream following manufacturer standard protocol. The Fit3D children’s body composition equation was used to derive fat mass (FM). Percent fat (%BF) was derived from dividing fat mass by known scale weight. Results from each system were compared to our criterion method, dual energy X-ray absorptiometry (DXA), with linear regression. Stepwise linear regression was used to derive scanner specific equations.
One hundred forty-nine participants were used in this analysis. Each participant needed a valid DXA, Styku, and Size Stream scan. Using the Fit3D derived equations, Styku and Size Stream had high correlations to DXA total fat (R2 = 0.90 and 0.85, RMSE = 6.38 kg and 28.11 kg, all respectively). System specific derived equations had a higher correlation to DXA total fat (R2 = 0.93 and 0.96, RMSE = 3.82 kg and 2.18 kg, respectively).
The high correlation and lower RMSE with the system specific equations support that 3DO scanners need specific equations derived with their own variables to accurately predict body composition.