3D Scanning For Obesity First Results From The ADEPS Project

Willem DE KEYZER, University College Ghent, Belgium
DERUYCK F. 2 , VAN DER SMISSEN B. 3 , VASILE S. 4 , COOLS J. 4 , DE RAEVE A. 4 , DE HENAUW S. 5, 1 , CUYPERS K. 6 , HUYBRECHTS I. 7 , VAN RANSBEECK P. 3

1 Bio- and food sciences, University College Ghent, Ghent, Belgium
2 Exact sciences, University College Ghent, Ghent, Belgium
3 Mechatronics, University College Ghent, Ghent, Belgium
4 Fashion, textile and wood technology, University College Ghent, Ghent, Belgium
5 Public health, Ghent University, Ghent, Belgium
6 Public Health and Surveillance, Scientific Institute of Public Health, Brussels, Belgium
7 Dietary Exposure Assessment Group, International Agency for Research on Cancer, Lyon, France

Purpose
Obesity is associated with an increased amount of adipose tissue and linked to increased risks for certain types of cancer. Determination of body fat percentage (%BF) is not always possible due to limitations in available resources. Therefore, weight indexes like BMI offer a major advantage because they are quick and inexpensive to use. Although the BMI is extensively used, it is unable to differentiate adipose tissue from lean body mass. Therefore, the principal aim of the ADEPS project is to examine the extent to which %BF can be predicted using anthropometric measurements and to develop predictive equations useful as field method to asses %BF in clinical practice and research.
Methods
A dataset of anthropometric measurements obtained by 3D body scanning (n=1200, males & females, 18-65 years) was available within the research unit. From these data, samples of candidate anthropometrical measurements for total body volume prediction were selected. Regression analysis on sequentially selected datasets yielded anthropometric predictors useful to create a predictive model for densitometry-based %BF estimation. This model was then validated by comparison with %BF obtained from air-displacement plethysmography in a validation sample (n=200).
 
Results
Correlations of anthropometrics with total body volume (L) of body scans were evaluated. The strongest model in females included waist girth, body height, thigh girth and wrist girth (r² = 0,96, RMSE = 1,46). In males the model included waist girth, body height, thigh girth and upper arm girth (r² = 0,98, RMSE = 1,58). The recruitment of participants for validation is ongoing.
 
Conclusions
Our analyses show that more than 96% of variation in body volume can be predicted using the selected anthropometric measurements. From this volume, %BF can be calculated using known densities for fat and fat-free mass. Accuracy of these %BF predictions will be presented during the conference.
 
Funding
University College Ghent