Outline of Proposed Research:
This research aims to develop a method for generating synthetic DXA (Dual-energy X-ray Absorptiometry) images using advanced 3D optical body scans, offering a safer and more accessible alternative for assessing body composition. DXA is the current gold standard for measuring fat, lean tissue, and bone density, but it involves radiation exposure, high cost, and limited availability.
The central question guiding this research is:
Can 3D body surface scans, combined with relevant demographic and physiological data, be used to accurately generate synthetic DXA images for body composition analysis?
The objectives of this project:
To train a machine learning model capable of translating 3D body scans into synthetic DXA-like images which requires pre-training on large scale DXA images.
To evaluate the accuracy of the synthetic images in replicating clinically relevant body composition metrics.
To explore how incorporating additional factors such as age, sex, ethnicity, height, weight, and functional performance enhances prediction accuracy.
The scientific rationale lies in the potential of 3D scanning technology and AI to simulate internal body composition from external morphology. Success in this area could reduce dependence on DXA equipment, lower healthcare costs, and broaden access to routine health monitoring.