•  
  •  
 

THE TEST-RETEST RELIABILITY OF BODY COMPOSITION MEASURED USING DIGITAL IMAGES FROM A SMARTPHONE APPLICATION

Abstract

Madeline L. Schwing, Casey J. Metoyer, Katherine Sullivan, Mary E. Lovelady, Michael R. Esco, FACSM, Michael V. Fedewa, FACSM. University of Alabama, Tuscaloosa, AL, AL.

BACKGROUND: The ability to measure and track changes in muscle and fat is important for practitioners in the Allied Health and Sports Performance fields. An automated image analysis program was recently developed to measure muscle and fat from a single digital image using a smartphone application. However, the reliability of the application has yet to be assessed. PURPOSE: The purpose of this study was to evaluate the test-retest reliability of %Fat estimates from a single digital image when measured on two consecutive days. METHODS: A convenience sample of participants were included in the study (n=12, 83.33% female, 83.33% Caucasian 31.25±10.49 yrs, 24.82 kg/m2). Data collection occurred on two consecutive days with no more than 36 hours between visits. On Day 1, age, gender, and race were assessed via self-report. A full-body image from the posterior view was taken using an iPad Air 2 (Apple Inc., Cupertino, CA) against a white photography backdrop. A light meter (MT-912, Shenzhen Flus Technology Co., Ltd., Shenzhen, China) was used to measure brightness in Lux and ensure that testing conditions were consistent across both days. Participants returned on Day 2 and performed a second %Fat measurement under similar lighting and backdrop conditions. Images were analyzed using an automated smartphone application (made Health and Fitness LLC, Birmingham, AL. version 1.1.3), which provided estimates of %Fat using a proprietary algorithm. A paired samples t-test was used to assess potential mean differences in %Fat across the two trials. The test-retest reliability across the trials was measured using Pearson’s r, and described as weak, moderate, strong, or near-perfect (r=0.2, 0.5, 0.8, or 0.9, respectively). Data are presented as mean± standard deviation, with statistical significance set at p<0.05. RESULTS: No significant mean differences were observed between measurements obtained on Day 1 (27.16±5.08 %Fat) and Day 2 (27.04±5.49 %Fat) (p=0.65). In addition, a near-perfect correlation was observed between the trials (r=0.99, p<0.001). CONCLUSION: Given the negligible difference between measures and the near-perfect correlation, an inexpensive and portable technique to measure %Fat in field settings may be a valuable alternative when traditional assessment techniques are not available. Future research should examine the reliability across multiple camera types, image resolutions, lighting conditions, and color backgrounds.

This document is currently not available here.

Share

COinS