3DBODY.TECH 2021 - Paper 21.43

J. A. Shepherd et al., "Evaluating the Accuracy of an Hallucinatory Algorithm to Predict Body Shape Changes from Dieting and Physical Activity", Proc. of 3DBODY.TECH 2021 - 12th Int. Conf. and Exh. on 3D Body Scanning and Processing Technologies, Lugano, Switzerland, 19-20 Oct. 2021, #43, https://doi.org/10.15221/21.43.

Title:

Evaluating the Accuracy of an Hallucinatory Algorithm to Predict Body Shape Changes from Dieting and Physical Activity

Authors:

John A. SHEPHERD 1, Michael WONG 2, Isaac TIAN 3, Yong En LIU 1, Samantha KENNEDY, 4 Dylan LOWE 5, Nisa KELLY 1, Julia MW. WONG 6, Cara B. EBBELING 6, David S. LUDWIG 6, Ethan WEISS 5, Brian CURLESS 3, Steve B. HEYMSFIELD 4

1 Department of Epidemiology and Population Sciences, University of Hawaii Cancer Center, Honolulu HI, USA;
2 Department of Nutrition Sciences, University of Hawaii Cancer Center, Honolulu HI, USA;
3 Department of Computer Science, University of Washington, Seattle WA, USA;
4 Pennington Biomedical Research Center, Baton Rouge LA, USA;
5 Cardiovascular Research Institute, University of California, San Francisco CA, USA;
6 New Balance Foundation Obesity Prevention Center, Boston Children’s Hospital and Department of Pediatrics, Harvard Medical School, Boston MA, USA

Abstract:

Background.
Visualizing body shape at some future time point can be useful in a variety of ways including motivational support for health interventions where fat and muscle mass changes are anticipated to create positive body image reinforcement. Although there are a variety of "hallucinatory" algorithms available, most, if not all, are created using cross-sectional population modeling. Previous work in our group has shown that we can accurately predict changes in total body fat and lean mass using a baseline model applied to baseline and follow-up three-dimensional optical 3DO scans using principal components (PC) of the scan mesh points as predictor variables. In this study, we explore the spatial accuracy of hallucinated and actual scans after diet and physical activity interventions.

Methods.
Participants were recruited for a "shape model" from the Shape Up! Adult Study stratified by sex, BMI, age, and five ethnicities. All participants received 3DO scan using the Proscanner (Fit3D Inc., San Mateo, CA, USA) and total body dual energy X-ray absorptiometry (TBDXA) scans (Hologic Inc., Marlborough, MA, USA). The 3DO scans were co-registered to a standard 110,000-point mesh (Meshcapade GmbH, Tubingen, Germany) and meshes for all participants were transformed to a PC space to orthogonalize and reduce the dimensions of the data. Target features such as DXA and demographic variables (i.e., height, weight, fat mass, lean mass) were applied to the PC weights to create the manifold regression matrix. Once the coefficients have been found for these features, individual scans can be substituted into the equation to solve for the best estimate of an individual's 3DO scan at different feature values (i.e., fat mass, lean mass, weight, etc.) This shape model was applied to a second population of participants from three intervention studies: Shape Up!, FB4, Louisiana State University Athlete's Studies. Hallucinated 3DO scans were created using the known changes in DXA fat and lean mass and these scans were compared to the actual scans taken after the interventions. Hallucinated scans were subtracted from the actual intervention scan on a point-by-point basis. The point location differences for all participants were represented as a second PC model to represent the modes of variance between the hallucinations and actual scans.

Results.
There were 377 adults (167 male) in our shape model and 106 participants (67 male) in our intervention dataset. The first 15 PC shape model components describe 99% of the variance. Hallucination scans for one participant are shown in Figure 1 representing changes of +/-20kg of fat and +/-20kg of lean. The hallucination variances for one male and female participant in the intervention group are shown in Figure 2 as a color heat map.

Conclusion.
We present a method to visualize the accuracy of hallucinated scans that can be used in a general way to study how the human body changes shape with body composition changes.

Details:

Paper: 2143shepherd.pdf
Proceedings: 3DBODY.TECH 2021, 19-20 Oct. 2021, Lugano, Switzerland
Paper id#: 43
DOI: 10.15221/21.43
Presentation video: 3DBodyTech2021_43_Shepherd.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.