Comparison of automated post-processing techniques for measurement of body surface area from 3D photonic scans

摘要

Chuang-Yuan Chiu$1David L. Pease$2Samantha Fawkner$3Marcus Dunn$1Ross H. Sanders$4 $1Centre for Sports Engineering Research, Sheffield Hallam University, Sheffield, UK$2Australian Institute of Sport, Movement Science, Canberra, Australia$3lnstitute for Sport, Physical Education & Health Sciences, The University of Edinburgh, Edinburgh, UK$4Faculty of Health Sciences, Discipline of Exercise and Sport Science, The University of Sydney, Sydney, Australia; c.chiu@shu.ac.uk;;;; Body surface area (BSA) measurement is important in engineering and medicine fields to determine parameters for various applications. Three-dimensional scanning techniques may be used to acquire the BSA directly. Nevertheless, the raw data obtained from 3D scanning usually requires some manual postprocessing which is time-consuming and requires technical expertise. Automated post-processing of 3D scans enables expedient BSA calculation with minimal technical expertise. The purpose of this research was to compare the accuracy and reliability of three different automated post-processing techniques including Stitched Puppet (SP), Poisson surface reconstruction (PSR), and screened Poisson surface reconstruction (SPSR) using manual post-processing as the criterion. Twenty-nine participants were scanned twice, and raw data were processed with the manual operation and automated techniques to acquire BSAs separately. The reliability of BSAs acquired from these approaches was represented by the relative technical error of measurements (TEM). Pearson's regressions were applied to correct BSAs acquired from the automated techniques. The limits of agreement (LOA) were used to quantify the accuracy of BSAs acquired from the automated techniques and corrected by regression models. The reliability (relative TEM) of BSAs obtained from PSR, SPSR and SP were 0.32, 0.30 and 0.82% respectively. After removing bias with the regression models, the LOA for PSR, SPSR and SP were (-0.0134 m~2,0.0135 m~2), (-0.0130 m~2,0.0132 m~2), (-0.0573 m~2, 0.0572 m2) respectively. It is concluded that PSR and SPSR are good alternative approaches to manual postprocessing for applications that need reliable and accurate measurements of BSAs with large populations.