A machine learning approach relating 3D body scans to body composition in humans

Pleuss, James; Talty, Kevin; Morse, Steven; Kuiper, Patrick; Scioletti, Michael; Heymsfield, Steven B.; Thomas, Diana M.

doi:10.1038/s41430-018-0337-1

Cited by 32 publications

(43 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Shape features identified in this study characterise deviations in torso shape that exist within the sample data and are invariant to the effects of scale, location and orientation. The information used to characterise individuals in our study differed from that used in previous studies by Loffler-Wirth et al 9 and Pleuss et al 24 . In these studies, large numbers of simple measures, such as lengths and girths and their ratios, were extracted from 3D body scan data and normalised with respect to height.…”

Section: Discussionmentioning

confidence: 86%

“…Though the size of the participant samples used in these studies were larger than in our study, the PCA procedure identified the same number of components to describe 95% of the variation present within the cohort. This suggests that shape information inherent within 3D scan data includes subtle variations requiring a greater number of principal components to describe them fully, as opposed to size measures which can be described in a smaller number of components 24 . Though it is currently unknown what all of the shape features captured in this study represents in terms of human health, these results further illustrate the wealth of information regarding body shape and weight distribution which cannot be captured by measurements used in current practice.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

How shape-based anthropometry can complement traditional anthropometric techniques: a cross-sectional study

Thelwell

Chiu

Bullas

et al. 2020

Sci Rep

View full text Add to dashboard Cite

Manual anthropometrics are used extensively in medical practice and epidemiological studies to assess an individual's health. However, traditional techniques reduce the complicated shape of human bodies to a series of simple size measurements and derived health indices, such as the body mass index (BMi), the waist-hip-ratio (WHR) and waist-by-height 0.5 ratio (WHT.5R). Three-dimensional (3D) imaging systems capture detailed and accurate measures of external human form and have the potential to surpass traditional measures in health applications. the aim of this study was to investigate how shape measurement can complement existing anthropometric techniques in the assessment of human form. Geometric morphometric methods and principal components analysis were used to extract independent, scale-invariant features of torso shape from 3D scans of 43 male participants. Linear regression analyses were conducted to determine whether novel shape measures can complement anthropometric indices when estimating waist skinfold thickness measures. Anthropometric indices currently used in practice explained up to 52.2% of variance in waist skinfold thickness, while a combined regression model using WHT.5R and shape measures explained 76.5% of variation. Measures of body shape provide additional information regarding external human form and can complement traditional measures currently used in anthropometric practice to estimate central adiposity. Measurements of size and shape of the human body are an important source of information for a range of scientific fields and applications. Traditional manual anthropometrics have been used extensively in medical practice and epidemiological studies to derive health risk indicators, since it has been suggested that human body shape is dependent on its underlying composition, including soft and skeletal tissues 1. Indices, such as the body mass index (BMI), waist girth and the waist-hip ratio (WHR) are used to assess variations in human body dimensions and physical health 2-4. Of these, BMI is most commonly used in current practice to determine the healthy weight range for individuals based on their height. However, BMI fails to distinguish between quantities of muscle and fat, which are of different density, and therefore is prone to misclassifying muscular individuals as being overweight or obese 5,6. Size measures, such as sagittal diameter, waist girth and WHT.5R have been found to demonstrate improved correlations with quantities of abdominal visceral fat and greater associations with metabolic disease risks compared to BMI 7,8. Relative measures, such as the WHR, provide information about the size of the abdomen relative to the rest of the body, so has been used as a proxy of torso shape and central obesity, defined as excess fat around the abdominal region 7. However, these relatively simple approaches to measuring external human form only utilise a small number of manual anthropometrics, which are prone to human error and limited by their simplicity, as they do not fully describe ...

show abstract

Section: Discussionmentioning

confidence: 86%

Section: Discussionmentioning

confidence: 99%

How shape-based anthropometry can complement traditional anthropometric techniques: a cross-sectional study

Thelwell

Chiu

Bullas

et al. 2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Body size and shape are governed by genetic and environmental factors, including lifestyle with potential impact for health. There is growing evidence that body shape and regional body composition are strong indicators of metabolic health 1,2 . For example, overweight and obesity increase risks for developing metabolic and cardiovascular diseases in an age-dependent manner 3 .…”

Section: Introductionmentioning

confidence: 99%

The aging human body shape

et al. 2020

View full text Add to dashboard Cite

Body shape and composition are heterogeneous among humans with possible impact for health. Anthropometric methods and data are needed to better describe the diversity of the human body in human populations, its age dependence, and associations with health risk. We applied whole-body laser scanning to a cohort of 8499 women and men of age 40-80 years within the frame of the LIFE (Leipzig Research Center for Civilization Diseases) study aimed at discovering health risk in a middle European urban population. Body scanning delivers multidimensional anthropometric data, which were further processed by machine learning to stratify the participants into body types. We here applied this body typing concept to describe the diversity of body shapes in an aging population and its association with physical activity and selected health and lifestyle factors. We find that aging results in similar reshaping of female and male bodies despite the large diversity of body types observed in the study. Slim body shapes remain slim and partly tend to become even more lean and fragile, while obese body shapes remain obese. Female body shapes change more strongly than male ones. The incidence of the different body types changes with characteristic Life Course trajectories. Physical activity is inversely related to the body mass index and decreases with age, while self-reported incidence for myocardial infarction shows overall the inverse trend. We discuss health risks factors in the context of body shape and its relation to obesity. Body typing opens options for personalized anthropometry to better estimate health risk in epidemiological research and future clinical applications.

show abstract

“…Concerning the method of selecting individual variables from the whole catalogue of scanned measurements, there are not many other studies we can use for comparison. Previous studies have aggregated meta-measures to cluster body types [44,45] or have used deep learning [43]. Moreover, recent research has also focused on developing methods using the 3D geometry of the surface topography in order to predict body shape [51,69,70].…”

Section: Plos Onementioning

confidence: 99%

“…As regards the potential use of multiple measures for a more precise prediction of body composition, previous studies only considered a limited number of predefined measures [42], and only few of them used automatic variable selection procedures to identify the best predictors [43][44][45]. Because some of the 150 standard measurements are strongly correlated among each other, model selection procedures and other techniques such as 3D surface geometry may have to account for these correlations [46][47][48][49][50][51]. Still, further research is needed to identify which of the 150 standard measurements are most relevant for the prediction of body composition, or whether multiple (and partly strongly correlated) measurements are relevant, and how they should be selected or combined to obtain the most reliable predictions.…”

Section: Introductionmentioning

confidence: 99%

Multiple measures derived from 3D photonic body scans improve predictions of fat and muscle mass in young Swiss men

et al. 2020

View full text Add to dashboard Cite

Introduction Digital tools like 3D laser-based photonic scanners, which can assess external anthropometric measurements for population based studies, and predict body composition, are gaining in importance. Here we focus on a) systematic deviation between manually determined and scanned standard measurements, b) differences regarding the strength of association between these standard measurements and body composition, and c) improving these predictions of body composition by considering additional scan measurements. Methods We analysed 104 men aged 19-23. Bioelectrical Impedance Analysis was used to estimate whole body fat mass, visceral fat mass and skeletal muscle mass (SMM). For the 3D body scans, an Anthroscan VITUSbodyscan was used to automatically obtain 90 body shape measurements. Manual anthropometric measurements (height, weight, waist circumference) were also taken. Results Scanned and manually measured height, waist circumference, waist-to-height-ratio, and BMI were strongly correlated (Spearman Rho>0.96), however we also found systematic differences. When these variables were used to predict body fat or muscle mass, explained variation and prediction standard errors were similar between scanned and manual measurements. The univariable predictions performed well for both visceral fat (r 2 up to 0.92) and absolute fat mass (AFM, r 2 up to 0.87) but not for SMM (r 2 up to 0.54). Of the 90 body scanner measures used in the multivariable prediction models, belly circumference and middle hip circumference were the most important predictors of body fat content. Stepwise forward model selection using the AIC criterion showed that the best predictive power (r 2 up to 0.99) was achieved with models including 49 scanner measurements.

show abstract

A machine learning approach relating 3D body scans to body composition in humans

Cited by 32 publications

References 19 publications

How shape-based anthropometry can complement traditional anthropometric techniques: a cross-sectional study

How shape-based anthropometry can complement traditional anthropometric techniques: a cross-sectional study

The aging human body shape

Multiple measures derived from 3D photonic body scans improve predictions of fat and muscle mass in young Swiss men

Contact Info

Product

Resources

About