Digital Food Photography: Dietary Surveillance and Beyond 2211-601X © 2013 The Authors. Published by Elsevier Ltd. Selection and peer-review under responsibility of National Nutrient Databank Conference Steering Committee doi: 10.1016/j.profoo.2013.04.019 Procedia Food Science 2 ( 2013 ) 122 – 128 36th National Nutrient Databank Conference Digital Food Photography: dietary surveillance and beyond Noemi G Islama*, Hafza Dadabhoya, Adam Gilluma, Janice Baranowskia, Thea Zimmermanb, Amy F Subarc, Tom Baranowskia aBaylor College of Medicine, Houston, TX, 77030, USA bWestat, Rockville, MD 20850, USA cNational Cancer Institute, Bethesda, MD 20892, USA Abstract The method used for creating a database of approximately 20,000 digital images of multiple portion sizes of foods linked to the for Dietary Studies (FNDDS) is presented. The creation of this database began in 2002 and its development has spanned 10 years. Initially the images were intended to be used as a kid-friendly aid for estimating portion size in the context of a computerized 24-hour dietary recall for 8-15 year old children. In 2006, Baylor College of Medicine, Westat, and the National Cancer Institute initiated a collaboration that resulted in the expansion of this image database in preparation for the release of the web-based Automated Self- Administered 24 Hour Dietary Recall (ASA24) for adults (now also available for use by children ASA24-Kids). Researchers in the US and overseas have capitalized on these digital images for purposes including, but not limited, to dietary assessment. © 2013 Published by Elsevier Ltd. Selection and peer-review under responsibility of the National Nutrient Databank Conference Editorial Board Keywords: digital images; portion size assessment; dietary surveillance; technology aided dietary assessment 1. Introduction The Food Intake Recording Software System (FIRSSt) [1] was originally developed as a computer- based, self-assessment tool to collect dietary information on number of servings of fruit and vegetables among school-age children. The success of this program, encouraged our team of nutritionists at the nutrient database and incorporating one of the top eleven technologies of the previous decade: digital photography [2] The idea was that the incorporation of digitally produced portion size images could * Corresponding author:. Tel.: +0- 713-798-7037; fax: +0-713-798-7130. E-mail address: nislam@bcm.edu Available online at www.sciencedirect.com © 2013 The Authors. Published by Elsevier Ltd. Selection and peer-review under responsibility of National Nutrient Databank Conference Steering Committee Open access under CC BY-NC-ND license. Open access under CC BY-NC-ND license. http://creativecommons.org/licenses/by-nc-nd/4.0/ http://creativecommons.org/licenses/by-nc-nd/4.0/ 123 Noemi G. Islam et al. / Procedia Food Science 2 ( 2013 ) 122 – 128 n. However, to produce a large number of standardized digital images with minimal effort, a standardized tool was needed and the Food Photography System (FPS) was such tool. The FPS made it possible to produce approximately 4000 digital images representing the food types, forms and range of portion sizes frequently reported by children in the Continuing Survey of Food Intake by Individuals 1994-1996-1998 (CSFII 94-96, 98). In February 2006, the Exposure Biology Program, one of the components of the Genes, Environment and Health Initiative (GEI) was launched, thus providing the first funding opportunity to move the food photography project forward. However, it was the partnership with the National Cancer Institute and Westat, which permitted the expansion of the digital image database to accommodate the requirements of a self-administered tool for both adults and kids, the web-based Automated Self -Administered 24 Hour Dietary Recall (ASA24) and ASA24-Kids [3]. 2. Methods 2.1. Food Photography System (FPS) The main goal of the FPS was to produce high quality, high resolution digital images with minimal technical expertise and limited budget. To achieve this goal, the system had to be able to meet the following requirements: Produce properly exposed images with sharp detail Allow to easily obtain aerial and angled (42º) exposures of the foods Be simple to operate: require minimal manipulation of the cameras with only sporadic calibrations by a professional photographer Allow control of the exposures Allow the storage and transfer of images to and from dedicated PCs without manipulation of the secure digital card (SD) 2.2. System design and components A front view of the main components of the FPS is shown in Figure 1 and the two angles of exposure are illustrated in Figure 2. Of the two cameras shown in Figure 2, the one at the top captures aerial views, is placed at a distance of 86.36 cm (measured from the focal point to the film plate) and is set at an angle of 5º from the vertical plane to eliminate reflections of the camera on the images. The second camera, placed on the left side, captures angled views at a distance of 91.44 cm (measured from the focal point to the film plate) and is set at an angle of 42º above the horizontal, considered to be the average angle of viewing for a subject seated at a dining table [4]. 124 Noemi G. Islam et al. / Procedia Food Science 2 ( 2013 ) 122 – 128 Fig.1. FPS system components Fig.2. Side view of the FPS showing camera angles 125 Noemi G. Islam et al. / Procedia Food Science 2 ( 2013 ) 122 – 128 2.3. Lighting and background All foods are photographed in a light booth constructed upon a kitchen preparation table. The booth is large enough to allow the placement and retrieval of the plate/bowl without disturbing the set-up. The booth dimensions are: 138 cm width, by 122 cm height and 91.8 cm depth. Light is supplied by two flash heads with integrated modeling lamps that provide continuous lighting for composing the image. The main light is a large light box suspended above the place setting and slightly behind the top camera. The light emanating from the lamp is softened through a layer of diffusion material that forms the top of the booth. Fill light comes from an umbrella reflector positioned under the table. The left and right sides of the booth are made of white matte board. A stand supports a paper backdrop that gradually curves from the back of the booth to cover the table surface. It is coated with Chroma-key blue paint, thus allowing graphic artists to easily drop out the background of the picture and replace it with one of their own. 2.4. Image capture and storage Each camera connects to its own PC via a USB cable. The software installed on each of the PCs controls the cameras remotely, enabling the viewing of the LCD monitor. The monitor display allows the operator to view the images as soon as the shutter is released. The images captured by the top and side cameras are stored on different PCs. Upon completion of the photo capturing session, these images are transferred to a network location for safe-keeping and image cataloguing. 2.5. Image quality and standardization Figure 3 shows a thumbnail size picture representing the aerial and angled view exposures of a ¾ cup of cheese-filled manicotti generated by the FPS. Both cameras are set to capture JPEG files of the highest quality and size (3 megabytes each). Images are extremely detailed and can be used for all types of output ranging from screen viewing to quality printing. End users have the flexibility to choose which way they want to process the images for their own specific needs. Image standardization in the FPS is achieved by comparing images obtained against reference images of the place setting captured and stored on the computer hard drives for calibration purposes. Additionally, minute registration marks are strategically placed on the backdrop to facilitate the realignment of the plate and/or cutlery if a displacement occurs when the plates with food are moved in and out of the booth during a photo capturing session. Fig.3. Aerial and angled views of ¾ cup of cheese filled manicotti with tomato sauce 126 Noemi G. Islam et al. / Procedia Food Science 2 ( 2013 ) 122 – 128 Another way to achieve standardization is to use the same type of plates and bowls. Baylor has consistently used Corelle® -oz Corelle® Winter Frost soup and dessert bowls (6.25 top dia). To avoid disturbing the place setting during image capturing, we use a base plate of the same type and size as a place holder. 2.6. System operation A system like the FPS requires minimum technical skill to be operated successfully. However, tasks such as food selection, acquisition, preparation, weighing and/or measuring of the food items to be depicted on the images according to a detailed protocol are better done by a professional with a background in food and nutrition (dietitians, dietetic technicians, or advanced food science students are the best operators). 3. Results and Discussion To date, the FPS has generated about 20,000 standardized portion size images of foods, in both aerial and angled views. Half of the images produced have been incorporated in the ASA24 web-based application [5]. The protocol used to capture portion size images for ASA24 required that all images be linked to an eight-digit FNDDS code and to a portion size descriptor and gram weight value. This feature sets the image database apart from others, and has resulted in a number of collaborations with researchers both in the US and abroad. Examples of this collaborative work include: A researcher from Denmark, who included a subset of our digital images to be used as portion size aids in a self-report tool for 9-11 year old Danish children, the Web-based Dietary Assessment Software for Children known as WebDASC [6] A researcher from Pennington Biomedical Research Center, LSU, who is using the images produced by the FPS as a standard or reference against which to compare before and after pictures captured by study participants using a smart phone [7] A researcher from the University of Washington, who used a subset of the digital images in video games for diabetes nutrition education A researcher from Indiana State University who used our standardized images to appropriately scale the digital images they had produced in their lab In addition to the above mentioned research applications, portion size food photography could play a role in interventions aimed at helping consumers better cope with an environment known to promote tion size images alongside nutrition information, could potentially have a positive impact, by providing a normative benchmark [10] of how much is appropriate package To create and maintain an image database of the proportions above described requires a considerable investment of time and resources. By making the image database freely available to other scientists and/or educators that request the permission to use it by contacting the National Cancer Institute (http://riskfactor.cancer.gov/tools/instruments/asa24/resources/contact.html) we seek to have the best possible return for such an investment. 127 Noemi G. Islam et al. / Procedia Food Science 2 ( 2013 ) 122 – 128 Fig 4.Nutrition facts: using portion size photography as a visual cue Fat 0.2 Calories 156 Carb 39 Protein 2.6 128 Noemi G. Islam et al. / Procedia Food Science 2 ( 2013 ) 122 – 128 References [1] Baranowski T, Islam N, Baranowski J, Cullen KW, Myres D, Marsh T et al. The Food Intake Recording Software System is valid among fourth-grade children. J Am Diet Assoc 2002;102:380 5. [2] Ross, PE. Top 11 technologies of the decade. Spectrum, IEEE 2011;48-1:27-63. [3] Subar AF, Kirkpatrick, BM, Mittl B, Zimmerman TP Thompson, FE, Bingley C et al. The Automated Self-Administered 24- Hour Dietary Recall (ASA24): A Resource for Researchers, Clinicians, and Educators from the National Cancer Institute. J Acad Nutr Diet 2012;112: 1134-7 [4] Nelson M, Atkinson M, Darbyshire S. Food photography I: the perception of food portion size from photographs. Br Med J 1994;72:649-663 [5] Zimmerman TP, Hull SG, McNutt S, Mittl B, Islam N, Guenther P et al. Challenges in converting an interviewer- administered food probe database to self-administration in the National Cancer Institute automated self-administered 24-hour recall (ASA24). J Food Comps Anal 2009;225:S48-S51 [6] Biloft-Jensen A, Christensen T, Islam N, Andersen LF, Egenfeldt-Nielsen et al. WebDASC: a web-based dietary assessment software for 8-11-year old Danish children. J Hum Nutr Diet 2012; DOI:10.1111/j.1365-277X.2012.01257.x [7] Martin CK, Han H, Coulon SM, Allen RH, Champagne CM, Anton, SD. A novel method to remotely measure food intake of free-living individuals in real time: the remote food photography method. Br Med J 2009;101:446-456 [8] Steenhuis IHM, Vermeer WM. Portion size: review and framework for interventions.International Journal of Behavioral Nutrition and Physical Activity 2009;DOI:10.1186/1479-5868-6-58 [9] EUFIC (2011). Consumer response to portion information on food and drink packaging A pan-European study. EUFIC Forum nº 5 [10] Wansink B. Environmental Factors that Increase the Food Intake and Consumption Volume of Unknowing Consumers. Annual Review of Nutrition 2004;24: 455-479 Presented at NNDC (March 25-28, 2012