3DBODY.TECH 2021 - Extended abstract 21.35

L. Leong et al., "Creating Accurate Representations of DXA Scans from 3D Optical Body Surface Scans for Arbitrary Regional Body Composition Analysis", Proc. of 3DBODY.TECH 2021 - 12th Int. Conf. and Exh. on 3D Body Scanning and Processing Technologies, Lugano, Switzerland, 19-20 Oct. 2021, #35, https://doi.org/10.15221/21.35.


Creating Accurate Representations of DXA Scans from 3D Optical Body Surface Scans for Arbitrary Regional Body Composition Analysis


Lambert LEONG 1,2, Michael WONG 1, Yong En LIU 1, Nisa N. KELLY 1, Michaela PIAZZA 3, Siobhan GARRY 4, Steve B. HEYMSFIELD 5, John A. SHEPHERD 1

1 University of Hawaii Cancer Center, Honolulu HI, USA;
2 University of Hawaii at Manoa, Honolulu HI, USA
3 National Institute of Allergy and Infectious Diseases, Bethesda MD, USA;
4 Hologic, Inc., Santa Clara CA, USA;
5 Pennington Biomedical Research Center, Baton Rouge LA, USA


Dual energy X-ray absorptiometry (DXA) has long been the gold standard method of quantifying the fat, lean, and bone composition of the body. Accessibility to DXA is a concern and not everyone is afforded the opportunity to receive a DXA scan. Three-dimensional (3D) body scanning technology offers an accessible method for accurately capturing the 3D surface of an individual. Previous works have demonstrated strong correlations between 3D body scans and metabolic risk factors as well as DXA comparable body composition. In essence, body shape, as measured by 3D body scanning technology, is highly correlated to and is a product of the underlying boney structures and soft tissue. In this work we derive mappings from 3D body scans to DXA scans to produce a model that can predict DXA equivalent images from a 3D body scan. We call this method Pseudo-DXA and generated images are compatible with DXA analysis software which can be used for arbitrary exploration of subregional body composition.

Participants recruited for the Shape Up! Study received whole body DXA scans (Hologic Inc., Marlborough, MA, USA) and 3D body scan on a Proscanner (Fit3D Inc., San Mateo, CA, USA). Principal component analysis (PCA) was used to generate four separate PCA models: three appearance models from the separate fat, lean tissue, and bone thicknesses, and one model to describe the variance in the 3D optical scan meshes. The end result was four sets of orthogonal and dimensionally efficient descriptions of the variance of fat, lean, bone, and 3D body shape. Linear regression was used to map the 3D optical model to each DXA PC model. Predicted DXA PC variables were then converted back to real space to create best estimate DXA images. Predicted DXA images were evaluated against the actual acquired DXA image in a hold-out set of images.

In total, 300 participants (150 male and 150 female) were available for this study. Twenty-seven 3DO PCs were used to 99% of the variance of the DXA fat images. Likewise, 31 3DO PCs were used to explain 99% of the variance in the DXA bone and lean images. Pseudo-DXA was able to predict DXA fat composition from a 3D body scan with a pixel root mean squared error (RMSE) of 15 mg on a female participant, see Figure.

We demonstrate a proof-of-concept method for predicting accurate, DXA equivalent, bone and soft tissue distributions from a 3D body scan. Pseudo-DXA may provide a more accessible method for evaluating arbitrary body regions for body composition important for many fields of study. Pseudo-DXA image can also be combined with the 3D body scans for visualization of soft tissue and joint centers which may improve safety and comfort when designing garments and protective wear.


3D scanning, foot scanning, footwear, mobile device


Extended abstract: 2135leong-eabs.pdf
Proceedings: 3DBODY.TECH 2021, 19-20 Oct. 2021, Lugano, Switzerland
Paper id#: 35
DOI: 10.15221/21.35
Presentation video: 3DBodyTech2021_35_Leong.mp4

Copyright notice

© Hometrica Consulting - Dr. Nicola D'Apuzzo, Switzerland, hometrica.ch.
Reproduction of the proceedings or any parts thereof (excluding short quotations for the use in the preparation of reviews and technical and scientific papers) may be made only after obtaining the specific approval of the publisher. The papers appearing in the proceedings reflect the author's opinions. Their inclusion in these publications does not necessary constitute endorsement by the editor or by the publisher. Authors retain all rights to individual papers.

Note: click the + on the top left of the page to open/close the menu.