3D scans, deep-learning AI show promise for measuring body composition

Editor's Note

A new method of measuring body composition using deep learning models and 3D scans is accurately depicts the distribution of fat and muscle in various body types, according to findings published January 30 Nature Communications Medicine.  

Creating tying an accurate digital map of a person’s shape and body composition to a corresponding x-ray can provide important information about a person’s health status and risks. However, such information has not been available previously. The new method, developed by researchers at Pennington Biomedical Research Center in Louisiana, uses 3D body surface scans that are fed into a dual-energy X-ray absorptiometry, or DXA scanner, which is able to then measure a patient’s quantities of muscle, fat, and bone.

The team confirmed the scanner’s results with scales and measuring tape. They found that their method produced more accurate measurements of fat, lean muscle, and bone than the commercial software which is currently used in clinical settings.

The body scans came from the "Shape Up! Adults" study, of which Pennington Biomedical was a part. The scans were originally designed for the clothing industry, but are now being adapted for biomedical use.  A non-invasive, accurate body scan could be a great tool to clinicians in measuring patients’ health and declines of muscle mass and strength due to aging and immobility conditions like sarcopenia, researchers conclude.