UCSF

Despite tremendous progress in biological understanding of metabolic diseases and development of novel imaging techniques over the past several decades, clinical assessments and epidemiological studies for metabolic disease are largely limited to simple surrogate measurements such as body mass index. Recent advances in dual-energy X-ray absorptiometry (DXA) and 3D optical surface scanning have enabled new opportunities for integrative and clinically-accessible metabolic health assessment.

The focus of this dissertation was to develop novel 3D and DXA imaging methods for detailed metabolic risk assessment that are fast, safe, and accessible. This work centered on the hypothesis that detailed descriptors of body shape and compositional distribution of tissues throughout the body better reflect health risk than currently-used metrics.

I present derivation of high-resolution quantitative fat, lean, and bone images from DXA, and describe statistical appearance models of these images that can be used to accurately predict metabolic syndrome and diabetes status across ethnicities. I demonstrate clinically-viable body composition estimation from commercial 3D optical body scans using traditional anthropometric measurements, then improved accuracy and precision using more detailed 3D statistical shape models that efficiently capture 95% of body shape variance. Finally, I describe integrative methods including a combination DXA and bioelectrical impedance technique to provide rapid, accurate, and precise four-component (4C) body composition. 4C assessment is useful for monitoring of many metabolic conditions including over/de-hydration, malnutrition, obesity, and sarcopenia, and this technique enables practical implementation in the clinic.

These works collectively provide new tools to researchers, clinicians, and even individuals around the world to assess metabolic status and track, visualize, and predict personalized body changes towards improved health.