There are several key components of health, which can be monitored through means of body composition and shape analysis. Metrics such as fat mass index and fat free mass index are simple features of the human body that can aid our understanding in their overall health. How these are measured have been extensively studied throughout the years. Means such as DXA and bioimpedance can accurately measure body fat. While more recently, 3D optical imaging has been shown to be a promising alternative whilst providing benefits such as lower cost. The Shepherd Lab has demonstrated such findings by showing 3D optical imaging to track well with DXA. The project is ongoing and this thesis is the culmination of my progress relating to 3D optical imaging methods and testing for applications in microgravity. It provides an overview of my work and the starting point of much future effort.
Through experiments, I optimized the equipment to maximize the performance of sensors and our imaging system for use in a parabolic flight setup utilizing microgravity simulation protocols to improve the measurement of health metrics through both performance and speed. I showed the importance of settings such as frame rate, resolution, and noise and how the specs may look similar between sensors but by recording metrics such as fill rate, we can see large areas of performance differences. With regards to the imaging configuration, I documented the relationship between certain settings and the resulting 3d reconstruction using quantitative metrics in which the distance from the sensor to the object is of high importance. I created a system that can be used on a parabolic flight with microgravity and future testing further out in space. Several models for learning body fat using a graph neural network, extreme gradient boosted decision trees were trained, and I produced a system for the first automated Caesar dataset landmark placement using deep learning which may hopefully greatly accelerate any pipelines that use this for template registration utilizing a robustly trained version.
|Commitee:||Shepherd, John, Baek, Kyungim|
|School:||University of Hawai'i at Manoa|
|School Location:||United States -- Hawaii|
|Source:||MAI 82/1(E), Masters Abstracts International|
|Subjects:||Computer science, Artificial intelligence, Electrical engineering|
|Keywords:||Body composition, Deep learning, Depth sensors, Neural network, Parabolic flight, Point cloud|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be