The use of photographs as an adjunct to estimated food records in dietary assessment
Background: Image-assisted dietary assessment is a promising area of research trialling ways to reduce the limitations of traditional dietary assessments. Additional images can increase the quality of nutrition information gained from dietary records across a range of population groups. Some automated imaging devices also give researchers extra information in order to investigate the social and environmental contexts influencing an individual’s intake.Objective: The aim of this study was to examine whether two single images of a meal are beneficial as an adjunct to a poor-quality and good-quality estimated food record in order to improve the accuracy and precision of the nutrition information entered into dietary assessment software.Methods: Fifteen meals: 3 breakfasts, 3 lunches and 9 dinners, including a range of cultural meal preferences, were made by the student researcher. Ingredients were weighed and meals photographed from a bird's eye view and 45⁰ angle, with a fork as a fiducial marker. Poor-quality and good-quality estimated food records and associated photographs were created and given to ten data enterers, five of whom were experienced in data entry, and five inexperienced. Data enterers reviewed the estimated food records and entered their estimations from the food records and accompanying photographs into the dietary assessment software, ‘Kai-culator’. A focus group with four student dietitians, discussing the effectiveness of additional photographs in relation to nutrition information gained from estimated food records, was conducted. Comparisons were made between the nutrition information from the actual weights of the ingredients and estimations from the food records. Mean values were compared within eleven categories: energy, three macronutrients and seven food groups. Results: There appeared to be no difference between the poor-quality and good-quality estimated food records and associated photographs and estimating the mean intakes for each meal within each category. However, the results suggested the good-quality estimated food record produced less variability. In nine of the eleven categories, the good-quality estimated food record produced meals with lower standard deviation values. For 83% of the meals, there was no significant difference between the estimated food records. The experience level of the data enterers had no effect on the estimated nutrition information produced. Current student dietitians agreed that the photographs were beneficial in order to gain greater knowledge of foods eaten, portion sizes and cooking methods. Conclusion: The use of two photographs (birds eye and 45⁰ angle) as an adjunct to the estimated food records appear to have had a beneficial effect on the accuracy of nutrition information gained from the estimated food records. The additional images reduced the differences between the estimated food records and were shown to be especially useful for improving the information gained from the poor-quality estimated food records. Further research into the area of image-assisted dietary assessment is needed to increase the evidence base.
Advisor: Krebs, Jeremy; Parry-Strong, Amber; Hall, Rosemary
Degree Name: Master of Dietetics
Degree Discipline: Human Nutrition
Publisher: University of Otago
Keywords: food records; dietary assessment; image-assisted; photographs; photos
Research Type: Thesis