A pose-independent method for accurate and precise body composition from 3D optical scans

Area of Science:

• Biomedical Engineering
• Anthropometry
• Medical Imaging

Background:

• Accurate body composition assessment is crucial for health monitoring.
• Traditional methods like DXA involve ionizing radiation and may not be widely accessible.
• Three-dimensional optical (3DO) scanning offers a non-ionizing alternative for body shape capture.

Purpose of the Study:

• To evaluate if standardizing the pose of 3DO whole-body scans enhances body composition accuracy and precision.
• To determine the effectiveness of digitally re-posed 3DO scans across diverse populations.

Main Methods:

• Healthy adults (n=472) underwent 3DO and DXA scans.
• 3DO scan data were processed using principal component analysis (PCA) after digital re-posing to a standard template.
• Prediction models for body composition were developed using stepwise linear regression on 3DO principal components (PCs).

Main Results:

• Three PCs captured 95% of shape variance in both male and female datasets after re-posing.
• High accuracy was observed for fat mass (R²=0.91-0.94) and fat-free mass (R²=0.92-0.95) compared to DXA.
• Visceral fat mass prediction showed good accuracy (R²=0.77-0.79).

Conclusions:

• Digitally re-posing 3DO scans to a standardized pose significantly improves body composition assessment accuracy and precision.
• This re-posing technique provides a viable, non-ionizing alternative to DXA for body composition analysis.
• The developed models are suitable for clinical and non-clinical applications, especially where DXA is inaccessible or radiation is a concern.