Generative deep learning furthers the understanding of local distributions of fat and muscle on body shape and health using 3D surface scans

Authors

• Lambert T. Leong, University of Hawaii, Honolulu, HI
• Michael C. Wong, University of Hawaii Cancer Center, Honolulu, HI
• Yong E. Liu, University of Hawaii Cancer Center, Honolulu, HI
• Yannik Glaser, University of Hawaii, Honolulu, HI
• Brandon K. Quon, University of Hawaii Cancer Center, Honolulu, HI
• Nisa N. Kelly, University of Hawaii Cancer Center, Honolulu, HI
• Devon Cataldi, University of Hawaii Cancer Center, Honolulu, HI
• Peter Sadowski, University of Hawaii, Honolulu, HI
• Steven B. Heymsfield, Pennington Biomedical Research Center, Baton Rouge, LA
• John A. Shepherd, University of Hawaii, Honolulu, HI

Document Type

Article

Publication Date

1-30-2024

Publication Title

Communications Medicine

Abstract

Background: Body shape, an intuitive health indicator, is deterministically driven by body composition. We developed and validated a deep learning model that generates accurate dual-energy X-ray absorptiometry (DXA) scans from three-dimensional optical body scans (3DO), enabling compositional analysis of the whole body and specified subregions. Previous works on generative medical imaging models lack quantitative validation and only report quality metrics. Methods: Our model was self-supervised pretrained on two large clinical DXA datasets and fine-tuned using the Shape Up! Adults study dataset. Model-predicted scans from a holdout test set were evaluated using clinical commercial DXA software for compositional accuracy. Results: Predicted DXA scans achieve R2 of 0.73, 0.89, and 0.99 and RMSEs of 5.32, 6.56, and 4.15 kg for total fat mass (FM), fat-free mass (FFM), and total mass, respectively. Custom subregion analysis results in R2s of 0.70-0.89 for left and right thigh composition. We demonstrate the ability of models to produce quantitatively accurate visualizations of soft tissue and bone, confirming a strong relationship between body shape and composition. Conclusions: This work highlights the potential of generative models in medical imaging and reinforces the importance of quantitative validation for assessing their clinical utility.

First Page

1

Last Page

9

DOI

10.1038/s43856-024-00434-w

Creative Commons License

This work is licensed under a Creative Commons Attribution 4.0 International License.

Recommended Citation

Leong, Lambert T.; Wong, Michael C.; Liu, Yong E.; Glaser, Yannik; Quon, Brandon K.; Kelly, Nisa N.; Cataldi, Devon; Sadowski, Peter; Heymsfield, Steven B.; and Shepherd, John A., "Generative deep learning furthers the understanding of local distributions of fat and muscle on body shape and health using 3D surface scans" (2024). LSU-LCMC Cancer Center Faculty Publications. 179.
https://digitalscholar.lsuhsc.edu/llcc_facpubs/179
10.1038/s43856-024-00434-w