Generalizable Methodology for Measurement and Analysis of Nutritional Intake

Abstract

According to the American Cancer Society (ACS), in 2018, more than 17 million people were diagnosed with cancer, and over 9 million patients died of cancer. Recent studies show that diet and lifestyle were two of the most common risk factors for related diseases and cancer. In Epidemiologic studies, enhanced dietary measurement tools are used to collect and analyze nutritional data. Traditional nutritional measurement methods include manual measurements and self-reporting, which introduce problems such as: 1) misreporting, 2) human error, and 3) involuntarily change of dietary habit. Recent advancements in sensing technologies have allowed for the development of various 3D measurement techniques for different applications. However, epidemiologists are still using self-reporting systems rather than automated 3D measurement tools, which demonstrates that there is significant room for improvement in this field. The volumetric measurement of a 3D object requires an accurate depth calculation using reference-based or depth sensing estimation techniques. This dissertation is a study of the design, modeling, and integration of a 3D measurement system based on the structured light system (SLS) theory. The system was designed, developed, and evaluated in dietary assessment applications, such as cancer and diabetes, using both a customized 3D scanner and a commercial off the shelf (COTS) depth sensor. The participants were selected randomly to test the system in different use-cases. When compared to similar systems, the results showed an average increase in accuracy of 30% and a reduction in measurement time by more than a factor of three. Based on the results and feedback from the researchers and users, the use of low-cost depth sensors in 3D measurement have drastically improved the quality of automated nutritional analysis. In addition to dietary assessment applications, the system was further developed and tested in other epidemiological fields, such as wound care management, air quality monitoring, and environmental disease tracking.