Article Title



Casey J. Metoyer, Katherine Sullivan, Mary Lovelady, Jacob Broeckel, Madelyn Simmang, Andrew Fields, Michael Fedewa, Michael Esco, FACSM. University of Alabama, Tuscaloosa, AL.

BACKGROUND: Recently, a smartphone application has shown promise for its ability to accurately measure metrics of body composition from a 2D full-body image. However, the reliability of this technique has yet to be explored. PURPOSE: The aim of this study was to examine the reliability of relative adiposity (%Fat) estimates from a single digital image across five different color backgrounds. METHODS: A sample of adults were recruited for this study (n=12, 31.25±10.49 yrs, 166.86±7.21 cm, 69.44±11.77 kg, 83% women, 83% Caucasian). %Fat was estimated from full-body images taken from the posterior view (iPad Air 2, Apple Inc., Cupertino, CA) in front of different color backgrounds (%FatBLACK, %FatORANGE, %FatGREEN, %FatGREY, and %FatWHITE). %FatWHITE served as the reference condition. Images were analyzed using an automated smartphone application (made Health and Fitness LLC, Birmingham, AL). Potential differences between conditions were assessed using a repeated measures ANOVA. Reliability across the different conditions was determined using ICC. Data are reported as mean±standard deviation. Statistical significance was assessed using an alpha level of p<0.05. RESULTS: No significant mean difference was observed between %FatBLACK (31.02±8.87 %Fat), %FatORANGE (27.32±5.51 %Fat), %FatGREEN (27.83±5.42 %Fat), %FatGREY (27.90±4.98 %Fat) and the reference %FatWHITE condition (27.15±5.08 %Fat) (all p>.05). A moderate correlation was observed between %FatBLACK and %FatWHITE (ICC=0.45, p=0.04), whereas a strong correlation was observed between %FatORANGE, %FatGREEN, %FatGREY and the reference %FatWHITE condition (ICC=0.99, 0.97, and 0.97, respectively; all p<0.001). DISCUSSION: Based on the results of this study, the 2D image application displayed acceptable reliability across different color backgrounds. While the ICCs for each condition were statistically significant (p < 0.05), the reliability for the black condition was considered moderate compared to strong for the other conditions. Practically, the difference could be due to the types of clothing in conjunction with the color of the background, for example black clothing against a black background. Although, for the most accuracy, it is recommended that users perform the measurements under similar environmental conditions, it appears that images taken in front of various background colors are comparable given the acceptable reliability.

This document is currently not available here.