Smartphone camera based assessment of adiposity: a validation study

Abstract

Body composition is a key component of health in both individuals and populations, and excess adiposity is associated with an increased risk of developing chronic diseases. Body mass index (BMI) and other clinical or commercially available tools for quantifying body fat (BF) such as DXA, MRI, CT, and photonic scanners (3DPS) are often inaccurate, cost prohibitive, or cumbersome to use. The aim of the current study was to evaluate the performance of a novel automated computer vision method, visual body composition (VBC), that uses two-dimensional photographs captured via a conventional smartphone camera to estimate percentage total body fat (%BF). The VBC algorithm is based on a state-of-the-art convolutional neural network (CNN). The hypothesis is that VBC yields better accuracy than other consumer-grade fat measurements devices. 134 healthy adults ranging in age (21–76 years), sex (61.2% women), race (60.4% White; 23.9% Black), and body mass index (BMI, 18.5–51.6 kg/m2) were evaluated at two clinical sites (N = 64 at MGH, N = 70 at PBRC). Each participant had %BF measured with VBC, three consumer and two professional bioimpedance analysis (BIA) systems. The PBRC participants also had air displacement plethysmography (ADP) measured. %BF measured by dual-energy x-ray absorptiometry (DXA) was set as the reference against which all other %BF measurements were compared. To test our scientific hypothesis we run multiple, pair-wise Wilcoxon signed rank tests where we compare each competing measurement tool (VBC, BIA, …) with respect to the same ground-truth (DXA). Relative to DXA, VBC had the lowest mean absolute error and standard deviation (2.16 ± 1.54%) compared to all of the other evaluated methods (p < 0.05 for all comparisons). %BF measured by VBC also had good concordance with DXA (Lin’s concordance correlation coefficient, CCC: all 0.96; women 0.93; men 0.94), whereas BMI had very poor concordance (CCC: all 0.45; women 0.40; men 0.74). Bland-Altman analysis of VBC revealed the tightest limits of agreement (LOA) and absence of significant bias relative to DXA (bias −0.42%, R2 = 0.03; p = 0.062; LOA −5.5% to +4.7%), whereas all other evaluated methods had significant (p < 0.01) bias and wider limits of agreement. Bias in Bland-Altman analyses is defined as the discordance between the y = 0 axis and the regressed line computed from the data in the plot. In this first validation study of a novel, accessible, and easy-to-use system, VBC body fat estimates were accurate and without significant bias compared to DXA as the reference; VBC performance exceeded those of all other BIA and ADP methods evaluated. The wide availability of smartphones suggests that the VBC method for evaluating %BF could play an important role in quantifying adiposity levels in a wide range of settings.Trial registration: ClinicalTrials.gov Identifier: NCT04854421.

Digital Single-Image Smartphone Assessment of Total Body Fat and Abdominal Fat Using Machine Learning

“…Majmudar et al [ 26 ] evaluated the use of convolutional neural networking, a type of machine learning, for analysis of smartphone 2D digital images to estimate %fat from two standing body positions. Adult participants wore form-fitting clothing without jewelry, socks, or shoes and were photographed in frontal and back poses with arms extended from the torso and legs separated.…”

“…Regardless of approach, 2D DP methods have relatively good reproducibility and similarities in group comparisons of estimated and predicted body fat variables. Although the investigators reported no significance between mean predicted and DXA-determined %fat values, proportional bias (e.g., tendency for greater errors from 2D DP with increasing %fat values) emerged and limits general use of these models [ 25 , 26 ]. In contrast, the results of the present study found no proportional bias between 2D DP-predicted and DXA-measured FM.…”

“…Use of proprietary prediction equations that fail to designate the physical characteristics and body shape variables used as independent variables confound the interpretation of findings of 2D DP measurement precision (reproducibility), trends in bias, and wide LOA [ 24 , 25 , 26 ]. A potential limitation is the failure to reveal the inclusion of body weight, a significant predictor of %fat and FM, hence affecting the independence of digitally acquired and derived body surface components in estimation of body shape and composition [ 24 , 25 , 26 ]. The present study, which is an evolution of the first use of 2D DP with a smartphone camera [ 19 ], uses machine learning to determine silhouette characteristics to derive a precise and accurate model to predict FM independently of body weight.…”

“…Initial, and some contemporary, digital imaging methods use three-dimensional (3D) techniques with general access constrained by size and cumbersome equipment, immobility, and cost [ 21 , 22 , 23 ]. The current widespread public availability of smartphone cameras enables acquisition of two-dimensional (2D) digital image photographs that overcome the practical limitations of 3D digital imaging systems [ 19 , 24 , 25 , 26 ]. Contemporary use of digital imaging methods focuses solely on estimation of total body adiposity but offers an opportunity to assess localized fat in body regions such as the abdomen.…”

“…Computational machine learning is a mathematical tool increasingly used in medical research to develop algorithms designed to make accurate predictions from large data sets [ 27 ]. The application of machine learning to digital imaging analysis to estimate body fat is limited, with reports of poor reproducibility of body composition estimates, weak concordance with reference values, and proportional bias [ 24 , 25 , 26 ].…”