3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology
Abstract Body composition prediction from 3D optical imagery has previously been studied with linear algorithms. In this study, we present a novel application of deep 3D convolutional graph networks and nonlinear Gaussian process regression for human body shape parameterization and body composition...
Saved in:
| Main Authors: | , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-02-01
|
| Series: | npj Digital Medicine |
| Online Access: | https://doi.org/10.1038/s41746-025-01469-6 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1823861536713605120 |
|---|---|
| author | Isaac Y. Tian Jason Liu Michael C. Wong Nisa N. Kelly Yong E. Liu Andrea K. Garber Steven B. Heymsfield Brian Curless John A. Shepherd |
| author_facet | Isaac Y. Tian Jason Liu Michael C. Wong Nisa N. Kelly Yong E. Liu Andrea K. Garber Steven B. Heymsfield Brian Curless John A. Shepherd |
| author_sort | Isaac Y. Tian |
| collection | DOAJ |
| description | Abstract Body composition prediction from 3D optical imagery has previously been studied with linear algorithms. In this study, we present a novel application of deep 3D convolutional graph networks and nonlinear Gaussian process regression for human body shape parameterization and body composition estimation. We trained and tested linear and nonlinear models with ablation studies on a novel ensemble body shape dataset containing 4286 scans. Nonlinear GPR produced up to a 20% reduction in prediction error and up to a 30% increase in precision over linear regression for both sexes in 10 tested body composition variables. Deep shape features produced 6–8% reduction in prediction error over linear PCA features for males only, and a 4–14% reduction in precision error for both sexes. All coefficients of determination (R 2) for all predicted variables were above 0.86 and achieved lower estimation RMSEs than all previous work on 10 metrics of body composition. |
| format | Article |
| id | doaj-art-2776465c01d244cc96b327be3c02ab8d |
| institution | Kabale University |
| issn | 2398-6352 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | npj Digital Medicine |
| spelling | doaj-art-2776465c01d244cc96b327be3c02ab8d2025-02-09T12:55:42ZengNature Portfolionpj Digital Medicine2398-63522025-02-018111310.1038/s41746-025-01469-63D convolutional deep learning for nonlinear estimation of body composition from whole body morphologyIsaac Y. Tian0Jason Liu1Michael C. Wong2Nisa N. Kelly3Yong E. Liu4Andrea K. Garber5Steven B. Heymsfield6Brian Curless7John A. Shepherd8Paul G. Allen School of Computer Science and Engineering, University of WashingtonPaul G. Allen School of Computer Science and Engineering, University of WashingtonUniversity of Hawaii Cancer Center, University of Hawaii - ManoaUniversity of Hawaii Cancer Center, University of Hawaii - ManoaUniversity of Hawaii Cancer Center, University of Hawaii - ManoaUCSF School of Medicine, University of California—San FranciscoPennington Biomedical Research Center, Louisiana State UniversityPaul G. Allen School of Computer Science and Engineering, University of WashingtonUniversity of Hawaii Cancer Center, University of Hawaii - ManoaAbstract Body composition prediction from 3D optical imagery has previously been studied with linear algorithms. In this study, we present a novel application of deep 3D convolutional graph networks and nonlinear Gaussian process regression for human body shape parameterization and body composition estimation. We trained and tested linear and nonlinear models with ablation studies on a novel ensemble body shape dataset containing 4286 scans. Nonlinear GPR produced up to a 20% reduction in prediction error and up to a 30% increase in precision over linear regression for both sexes in 10 tested body composition variables. Deep shape features produced 6–8% reduction in prediction error over linear PCA features for males only, and a 4–14% reduction in precision error for both sexes. All coefficients of determination (R 2) for all predicted variables were above 0.86 and achieved lower estimation RMSEs than all previous work on 10 metrics of body composition.https://doi.org/10.1038/s41746-025-01469-6 |
| spellingShingle | Isaac Y. Tian Jason Liu Michael C. Wong Nisa N. Kelly Yong E. Liu Andrea K. Garber Steven B. Heymsfield Brian Curless John A. Shepherd 3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology npj Digital Medicine |
| title | 3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology |
| title_full | 3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology |
| title_fullStr | 3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology |
| title_full_unstemmed | 3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology |
| title_short | 3D convolutional deep learning for nonlinear estimation of body composition from whole body morphology |
| title_sort | 3d convolutional deep learning for nonlinear estimation of body composition from whole body morphology |
| url | https://doi.org/10.1038/s41746-025-01469-6 |
| work_keys_str_mv | AT isaacytian 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT jasonliu 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT michaelcwong 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT nisankelly 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT yongeliu 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT andreakgarber 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT stevenbheymsfield 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT briancurless 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology AT johnashepherd 3dconvolutionaldeeplearningfornonlinearestimationofbodycompositionfromwholebodymorphology |