Assessment of a handheld fluorescence imaging device as an aid for detection of food residues on processing surfaces

Food Control 59 (2016) 243e249

Contents lists available at ScienceDirect

Food Control journal homepage: www.elsevier.com/locate/foodcont

Assessment of a handheld fluorescence imaging device as an aid for detection of food residues on processing surfaces Colm D. Everard a, *, Moon S. Kim b, Hoyoung Lee b a

School of Biosystems Engineering, University College Dublin, Dublin 4, Ireland Environmental Microbial and Food Safety Laboratory, US Department of Agriculture, Agricultural Research Service, Beltsville Agricultural Research Center, Beltsville, MD 20705, USA b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 1 December 2014 Received in revised form 13 May 2015 Accepted 20 May 2015 Available online 21 May 2015

Contamination of food with pathogenic bacteria can lead to foodborne illnesses. Food processing surfaces can serve as a medium for cross-contamination if sanitization procedures are inadequate. Ensuring that food processing surfaces are correctly cleaned and sanitized is important in the food industry to reduce risks of foodborne illnesses and their related costs. A handheld fluorescence imaging device was assessed for detection of three types of food residues that have been associated with foodborne illness outbreaks, i.e. spinach leaf, milk, and bovine red meat, on two commonly used processing surfaces, i.e. high-density polyethylene and food grade stainless steel. Fluorescence excitation at 405 nm was supplied by 4  10 W light emitting diodes. Interchangeable optical filters were selected to optimise the contrast between the food residues and processing surfaces, using hyperspectral fluorescence imaging. The fluorescence imaging plus image analysis differentiated food residues from the processing surfaces more clearly than visual inspection in ambient lighting. This optical sensing device can be used to detect food fouling on food processing surfaces over relatively large areas, and has potential for use in the food industry as an aid for detection of specific food residues. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Fluorescence Contaminant Imaging Processing surfaces

1. Introduction Food fouling of processing surfaces is a major concern for the food industry, as it can lead to food safety issues (Barish & Goddard, 2014; De Jong, 1997). Accumulation of food residue can provide an environment for microbial growth and biofilm formation (Agle, 2007; Barish & Goddard, 2014; Parkar, Flint, & Brooks, 2004). Microorganisms can grow rapidly in food residues remaining on food processing or handling equipment after use (Jun et al., 2010). Controlling microorganisms is essential in food processing, in order to provide safe, wholesome, and palatable food to consumers (Hood & Zottola, 1995). Painter et al. (2013) estimated that more than 9 million foodborne illnesses are caused by pathogens each year. Cross-contamination with pathogenic bacteria such as Escherichia coli O157:H7, Salmonella enterica, and Listeria monocytogenes from food processing surfaces to food products can occur due to inadequate cleaning or sanitizing (Jun et al., 2010; Reij & Den Aantrekker, 2004). Processing such as trimming, cutting, washing,

* Corresponding author. E-mail address: [email protected] (C.D. Everard). http://dx.doi.org/10.1016/j.foodcont.2015.05.030 0956-7135/© 2015 Elsevier Ltd. All rights reserved.

rinsing, dewatering, and packaging are points of potential crosscontamination during fresh produce production (Srey, Jahid, & Ha, 2013). For example, Haeghebaert, Le Querrec, Vaillant, Delarocque Astagneau, and Bouvet (2001) suggested that 40% of the foodborne diseases caused by bacteria between 1996 and 1998 in France were related to contaminated equipment. Prolonging the shelf-life of fresh-cut produce can be achieved by washing, and inclusion of sanitizers in the wash solutions can reduce bacterial counts by as much as 2 log (Srey et al., 2013; Whipps, Hand, Pink, & Bending, 2008). However, foodborne pathogens such as L. monocytogenes can be difficult to eliminate, as they can survive extreme conditions of temperature, pH, and salts (Cole, Jones, & Holyoak, 1990; Koo, Ndahetuye, O’Bryan, Ricke, & Crandall, 2014). Diligent cleaning and sanitation inspection by restaurant owners, food suppliers, caterers, and others who handle and serve large volumes of food are necessary to reduce foodborne illnesses. Fresh produce has a high risk of association with foodborne illness because there is no lethal phase (e.g. heating to kill pathogens) before it is consumed (Wiederoder, Liu, Lefcourt, Kim, & Lo, 2013). Leafy greens have an enhanced risk of contamination in the field with pathogenic bacteria from fecal matter from livestock or wild animals that may enter the field (Everard, Kim, & Lee, 2014).

244

C.D. Everard et al. / Food Control 59 (2016) 243e249

Milk is vulnerable to contamination by microorganisms from ineffectively cleaned and sanitized equipment (Jessen & Lammert, 2003; Koutzayiotis, 1992; Srey et al., 2013). For example, the recall of approximately 20 pounds of raw milk in 2011 in Washington State, USA, due to L. monocytogenes contamination was thought to be associated with biofilm formation (Srey et al., 2013). Bacteria of the genera Enterobacter, Listeria, Lactobacillus, Micrococcus, Streptococcus, Bacillus, and Pseudomonas are frequently encountered in the dairy environment (Salo, Ehavald, Raaska, Vokk, & Wirtanen, 2006; Sharma & Anand, 2002; Waak, Tham, & Danielsson-Tham, 2002). Cross-contamination of meat products is a concern, because some cooking practices do not kill all pathogenic bacteria. Contamination can occur during the slaughtering, dressing, chilling, or cutting stages of processing (Dourou et al., 2011; Koutsoumanis & Sofos, 2004). Many reports demonstrate the potential for cross contamination with E. coli O157:H7 via surfaces of equipment used for beef processing (Aslam, Greer, Nattress, Gill, & McMullen, 2004; Gill & McGinnis, 2000; Gun, Yilmaz, Turker, Tanlasi, & Yilmaz, 2003). Biofilm formation is recognized as a frequent source of crosscontamination in the food industry, because it allows bacteria to resist cleaning and disinfection. The biofilm serves as a barrier to prevent or lessen contact with the disinfectant (O’Toole & Kaplan, 2000; Srey et al., 2013). Biofilms are usually composed of water, proteins, lipids, and polysaccharides, as well as the bacteria (Donlan & Costerton, 2002; Jun et al., 2010). Srey et al. (2013) described five steps of biofilm formation, i.e. (1) initial attachment, (2) irreversible attachment, (3) early development of biofilm architecture, (4) maturation, and (5) dispersion. The final step enables cross-contamination by releasing pathogens back into the surrounding environment (Silagyi, Kim, Lo, & Wei, 2009). The attachment and formation of biofilms containing pathogenic microorganisms in food residues on processing surfaces or equipment is of major concern to food processors, because it can lead to cross-contamination and has the potential for major heath and economic consequences (Dourou et al., 2011). The role of food processing surface material, plant design, cleaning procedures, and sanitizers in ensuring safe food production are widely reported (Hadjiev, Dimitrov, Martinov, & Sire, 2007; Le Gentil, Sylla, & Faille, 2010; Palmer, Flint, & Brooks, 2007). Many products are available for cleaning of food processing surfaces, including surfactants and alkali products (Srey et al., 2013). Popular types of disinfectants include chlorine, hydrogen peroxide, iodine, ozone, and peracetic acid (Chmielewski & Frank, 2007; Srey et al., 2013). The effectiveness of antimicrobial agents in killing microorganisms is reduced by the presence of organic food residues (Srey et al., 2013). Cleaning can be targeted to dissolve the extracellular polymeric substance (EPS) matrix of biofilms, which allows the disinfectants to kill the bacterial cells that were protected by the ~es, Simo ~es, Machado, Pereira, & Vieira, 2006; Srey matrix (Simo et al., 2013). Clean in place (CIP) is a commonly-used process in the food industry, whereby the processing system is cleaned without dismantling and without an operator (Srey et al., 2013). CIP procedures and their effectiveness have been reported (Boulangevre, Petermann, Jullien, Dubois, Benezech, & Faille, 2004; Lelie ne zech, 2002). Antonini, Faille, & Be An important step in food safety is to ensure that the cleaning and sanitizing procedures have been effective. Cleaning and sanitation inspections usually involve visual inspection, adenosine triphosphate (ATP) bioluminescence assays, and culturing techniques such as rapid PCR, to assess sanitation effectiveness and reduce cross-contamination (Moore & Griffith, 2002; Wiederoder et al., 2013).

The ATP bioluminescence assay is a widely used method to monitor food processing surfaces in the food industry (Davidson, Griffith, Peters, & Fielding, 1999; Koo et al., 2013). The results are available within a few minutes but the assay detects both microorganisms and food residues, which can lead to inconsistent correlation with the level of bacterial contamination (Aycicek, Oguz, & Karci, 2006; Koo et al., 2013). In contrast, rapid PCR cultures take rez-Rodríguez, Valero, Carrasco, 24e48 h to obtain results (Pe García, & Zurera, 2008). However, since DNA can remain intact for up to 3 weeks after cell death, this can lead to overestimation or false positives in the detection of living microorganisms (Martinon, Cronin, Quealy, Stapleton, & Wilkinson, 2012; Nocker, Cheung, & Camper, 2006). Another method which can provide accurate, real-time results over large areas is needed (Wiederoder et al., 2013). Non-destructive and non-contact optical techniques that can monitor large areas and rapidly detect anomalies have gained considerable interest in the food industry (Everard et al., 2014; Jun et al., 2010). Many organic compounds fluoresce in the visual and near infrared wavebands when exposed to ultraviolet or violet excitation. This property could be utilized to detect contamination on food processing surfaces. Chlorophyll a has a distinctive fluorescence emission profile, with peaks near 685 and 730 nm (Everard et al., 2014). Fluorescence emissions peaks for other plant constituents have been reported near 340, 450 and 530 nm (Corp, McMurtrey, Chappelle, Daughtry, & Kim, 1997; Kim, McMurtrey, Mulchi, Daughtry, Chappelle, & Chen, 2001). Milk components that fluoresce include aromatic amino acids, vitamin A, and riboflavin (Christensen, Becker, & Frederiksen, 2005). Processing of milk also forms fluorescent compounds, e.g. Maillard reaction products (Birlouez-Aragon et al., 1998; Birlouez-Aragon, Sabat, & Gouti, 2002). Meat products have high florescence emissions in the ultraviolet (UV) and blue-green regions of the spectrum; the UV emissions are related to protein, and the blueegreen emissions are associated with aromatic compounds (Wold & Kvaal, 2000; Wold, Lundby, & Egelandsdal, 1999). A handheld fluorescence imaging device could be a useful aid for detection of food residues which are not easily discernible by the human eye (Cho, Chen, & Kim, 2007). Additional anticipated benefits of fluorescence imaging would be that larger areas than possible by swab sampling techniques could be assessed, and in real-time. The objectives of this study were to assess the usefulness of a recently developed handheld fluorescence imaging device (HFID), engineered in-house, to detect food residues on typical types of food processing surfaces. Fluorescence emission profiles, captured using hyperspectral fluorescence imaging, were used to identify the fluorescence emissions from tested materials, and to select appropriate optical filters for differentiating the food residues from the processing surfaces. 2. Material and methods 2.1. Fouling of food processing surfaces with food residues Two widely used food processing surface materials were used in this study, i.e. 45 cm  45 cm  6 mm white high-density polyethylene (HDPE) sheets (The Cutting Board Factory, Carbondale, PA, USA) and 45 cm  45 cm food grade 304 stainless steel (SS) sheets (2B finish; Stainless Supply, Monroe, NC, USA). Plastic polymers such as HDPE are often used in the food industry, in the manufacture of conveyor belts (Pompermayer & Gaylarde, 2000) and as a cutting board material (Jun et al., 2010). HDPE fluoresces under violet light, whereas the SS is non-fluorescent (Jun et al., 2010). Stainless steel is the most frequently used material for food

C.D. Everard et al. / Food Control 59 (2016) 243e249

processing surfaces, and it is readily cleaned (Boulange-Petermann, 1996). Residues of three foods were assessed, i.e. leafy green produce represented by baby spinach leaves (Spinacia oleracea), dried fat-free milk, and bovine red meat. For the leafy green residue, the spinach leaves were crushed, smeared on the food processing surface by hand, and allowed to dry (5 ± 0.5 h) at room temperature (20  C) before imaging. Spinach leaf residues which were clearly distinguishable from the processing surface by the human eye were not used in the trial. The area of each leaf residue sample was approximately 2 ± 1 cm2; 100 leaf residue samples were assessed on each processing surface type. For the fat-free milk residue, the milk was diluted to 1:1, 1:2, and 1:5 with distilled water, 40 samples (20 ml, approximately 5 mm diameter) at each dilution level were applied to each processing surface using a pipette (10e100 ml, Rainin LTS PipetPlus, MettlerToledo, Columbus, OH, USA), and allowed to dry (5 ± 0.5 h) at room temperature before imaging. Bovine red meat residues or chunks (5 ± 2 mm3) were also assessed on the processing surfaces; 100 meat chunk samples were evaluated on each processing surface. Meat samples did not leave a protein residue by smearing, therefore samples were cut and placed on the processing surfaces. Imaging was carried out 5 ± 0.5 h after preparation. Both the leafy green and fat-free milk residues had low visibility to the human eye or were difficult to distinguish from the background processing surfaces. The meat residues were distinguishable from the processing surfaces by the human eye at close inspection; however an optical sensing technology would prevent human error. 2.2. Hyperspectral fluorescence imaging A line-scan push-broom hyperspectral imaging (HSI) system, with recently developed light emitting diode (LED) based violet fluorescence excitation light sources, was used to assess the fluorescence of the food residues and processing surfaces in this study, to help determine which emission light filters would maximize the contrast between the residues and processing surfaces. The HSI system consisted of an electron-multiplying charge-coupled-device (EMCCD), a camera (MegaLuca R EMCCD, Andor Technology

245

PLC, Belfast, Northern Ireland), an imaging spectrograph (Hyperspec VS, Headwall Photonics, Fitchburg, MA, USA) with a 60 mm slit, a lens (Schneider-Kreuznach Xenoplan 1.4/23 C-mount, Schneider Optics, Hauppauge, NY, USA), and a motorized positioning table (Velmex, Bloomfield, NY, USA). Two line lights (built in-house), each with four 405 nm 10 W LEDs (LedEngin, CA, USA), provided violet fluorescence excitation. Incremental steps and exposure times were selected to provide optimal fluorescence over the spectral region of interest (data not presented). Fluorescence spectra were collected from 464 to 800 nm at intervals of approximately 4 nm, giving a total of 85 wavebands. Samples of the food residue materials were placed on non-fluorescent black background plates during HSI image acquisition. Image acquisition and analysis used software developed in Visual Basic (Version 6, Microsoft, Seattle, Washington, USA). The HSI system was also used to determine the spectral profile of the fluorescent ceiling lights in the trial room (ambient lighting). 2.3. Handheld fluorescence imaging device The in-house built HFID which was assessed here as an aid to detect food residue on HDPE and SS processing surfaces was described by Lee, Kim, Chao, Lefcourt, and Chan (2013). The main components include the excitation light source, provided by four 405 nm 10 W LEDs (Metal Core Printed Circuit Board with four embedded dice, LZ4-00UA00, LedEngin, CA, USA), interchangeable optical filters (470, 515, 640 and 680 nm with 10 nm bandwidths, TFI Technologies, Inc., Greenfield, MA, USA), and a charge-coupled device (CCD) camera with high quantum efficiency in the spectral range 400e700 nm (ICX424AL, Sony, Tokyo, Japan), all housed within a handheld casing (Fig. 1). In this study, relatively narrow emission bands (approximately 10 nm full-width at half maximum) were used to maximize the contrast between the food residues and the food processing surfaces. The CCD camera supported progressive scan, had a square pixel array (8.47 mm image), and EIA B/W output. The digitized image was transformed into wireless Ethernet data frames, using a video decoder (Model TVP5150AM1, Grandstream, Boston, MA), digital signal processor (Model

Fig. 1. Handheld fluorescence imaging device (Lee et al., 2013).

246

C.D. Everard et al. / Food Control 59 (2016) 243e249

processes, was removed from the fluorescence images using the “desparkle” function, prior to applying a threshold to distinguish the food residue from the food processing surface. The “desparkle” function used a median filter that replaced each pixel with the median value in its 3  3 neighbourhood. The masking threshold level was set in order to maximise the residue detection while eliminating the processing surface.

3. Results and discussion

Fig. 2. Handheld fluorescence imaging device set-up and field-of-view.

TMS320DM365, Grandstream, Boston, MA), and a video server (Model RTL8196C, Realtek, Hawthorne, CA, USA). This provided a real-time video stream to a PC located outside the light controlled trial room. Images were captured and saved in 8-bit grayscale JPEG format (704  480 pixels) for image processing. In addition to the above activities, images were also captured under ambient light only (i.e. HFID LEDs switched off), using the HFID with no emission filters in place and also with a high pixel digital camera (12.1 Mega Pixels; SX220 HS, PowerShot, Canon, Conns Cameras, Dublin, Ireland). The digital camera images were compared with the fluorescence images from the HFID. Fig. 2 shows the imaging set-up and field-of-view. 2.4. Image processing Image J (v.1.48; National Institutes of Health, USA) was used to analyse the captured images. Salt and pepper noise, which can result from the A/D conversion, Wi-Fi transmission, or 8-bit storage

▫

The HSI system and the HFID used the same LEDs for sample excitation, and had similar quantum efficiencies. This enabled the HSI emission profiles (Fig. 3) to be used to identify appropriate optical filters to use with the HFID. SS does not fluorescence. The wavelength at which fluorescence emission was highest for each food residue was used to identify appropriate optical filters for the HFID to maximize the contrast between the food residue and the SS processing surface. Fluorescence emission maxima were observed for spinach leaf residue, dried fat free milk and bovine red meat at 684, 516 and 615 nm, respectively. Since the HDPE processing surface did fluoresce, for food residue that had lower emissions than that of the HDPE surface, over the spectral range under investigation, the minimum emission wavelengths of that food residue was chosen to maximize the contrast. It was determined that bovine red meat had lower emissions over the measured spectral range than HDPE. The minimum fluorescence emission occurred at 464e470 nm for bovine red meat. The leaf residue on the HDPE surface was not clearly visible to the human eye (Fig. 4 i); only with close up inspection at a shallow viewing angle was it possible to observe a greenish hue. Similarly, in the image captured by the HFID with no optical filter, and under the trial room’s ambient light only, the leaf residue was not visible (Fig. 4 ii). However, the residue was clearly distinguished from the HDPE surface when the 405 nm fluorescence excitation source was used with a 680 nm optical filter in the HFID (Fig. 4 iii). This is consistent with the fluorescence emission of chlorophyll a molecules present in green plants. Chlorophyll a emits unique fluorescence in the red and far-red spectral regions; emission peaks at 685 and 730 nm have been reported (Everard et al., 2014; Kim, Lefcourt, & Chen, 2003; Kim et al., 2001).

Fig. 3. Typical fluorescence emission spectra profiles of spinach leaf residue ( ), dried fat-free milk (◊), bovine red meat (B) and HDPE (─ ─), excited at 405 nm. The spectra profile of the trial room’s ambient light (fluorescent ceiling lights) is also shown (──). Note that the exposure times for these profiles are different and therefore fluorescence emission intensities are not compared in this figure.

C.D. Everard et al. / Food Control 59 (2016) 243e249

Fig. 4. Typical examples of dry leaf residue on HDPE and SS processing surfaces.

Simple threshold determination and masking image analysis techniques (Fig. 4 iv, v, respectively) enhanced the distinction between the food residue and the HDPE surface. Fig. 4 vi shows the digital camera image of leaf residue on the SS processing surface under ambient light. The image was captured at an angle to prevent reflection of the overhead lights into the camera lens, i.e. scattered specular reflection was captured. Stainless steel has high specular reflectance; direct reflectance of the overhead lights (i.e. mirror image or shine) can interfere with detection of food residue. The leaf residue was visible under ambient light using the HFID without an optical filter (Fig. 4 vii; this image was also taken at an angle to avoid direct reflectance from the overhead lights). However, the spinach leaf residue was more clearly distinguished under fluorescence excitation and using a 680 nm optical filter (Fig. 4 viii, ix, x). 100% of the leaf residues were distinguished from both processing surfaces types using the HFID and appropriate conditions; Fig. 4 presents typical examples. An advantage of the HFID fluorescence image was the absence of direct reflectance, which impaired visual inspection. The fat-free milk residue on the HDPE surface was difficult to distinguish by human eye, because both have a white hue (Fig. 5 i). A 515 nm optical filter was used to maximize the contrast between the residue and the processing surface, for both the HDPE and SS, because the fluorescence emission of milk was relatively high at this wavelength with the 405 nm excitation source (Fig. 5 iii, viii). Vitamin A excitation and emission maxima occur at 327 and 510 nm, respectively (Ramanujam, 2000). The riboflavin emission peak is 510 nm (Liu & Metzger, 2007). The residue of fat-free milk was easily distinguished from the HDPE surface using the threshold and masking techniques (Fig. 5 iv, v). The fat-free milk residue was visible on the SS surface under ambient light at a camera angle which avoided direct reflection from the ceiling lights (Fig. 5 vi, vii). Under fluorescent excitation and using a 515 nm filter and image processing, the residue was clearly distinguishable from the SS surface (Fig. 5 viii), but the residue was much more clearly distinguishable after the threshold image processing and masking techniques were applied (Fig. 5 ix, x, respectively). All milk residue samples at each of the three dilution levels were distinguished in a similar manner using the HFID on both the HDPE and SS processing surfaces; Fig. 5 shows typical examples at the 1:5 dilution level.

247

Fig. 5. Typical examples of dry fat free milk (1:5 dilution) on HDPE and SS processing surfaces.

The chunks of bovine red meat were visible to the human eye on both HDPE and SS processing surfaces (Fig. 6 i, ii, vi, vii); however an inspection aid could be especially useful when there is a large processing surface area to be examined or when direct reflection (shine) from SS surfaces impairs visual inspection. An imaging device with real-time wireless display on a remote monitor would also enable access to surface areas not accessible to visual inspection. To maximize the contrast between the meat and the HDPE surface under fluorescence excitation, a 470 nm optical filter was used; at this wavelength the HDPE surface had higher fluorescence emissions than the meat chunks (Fig. 6 iii, iv, and v). For the SS surface, a 640 nm optical filter was selected. The meat fluorescence emission was relatively high at this wavelength, providing the

Fig. 6. Typical examples of bovine red meat on HDPE and SS processing surfaces (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.).

248

C.D. Everard et al. / Food Control 59 (2016) 243e249

Fig. 7. (i) Light intensity distribution over an image of bovine red meat samples on an HDPE processing surface, using the handheld fluorescence device; (ii) a surface plot of this image.

highest contrast against the non-fluorescent SS background (Fig. 6 viii, ix, and x). Light absorbance by the blood and fat components of the meat may have affected the fluorescence profile of the images. Although there is limited published data on hyperspectral fluorescence imaging of bovine red meet, colonization of beef processing surfaces by E coli O157:H7 has been reported (e.g. Aslam et al., 2004; Gun et al., 2003), emphasizing the importance of validation of cleaning procedures. All 100 meat residue samples on both the HDPE and SS surfaces were distinguished and typical examples are shown in Fig. 6. The arrangement of the LEDs and the CCD within the HFID device resulted in an uneven distribution of light exposure over the field-of-view of the instrument (Fig. 7). Since the HFID was originally designed to be passed manually over the surface area being measured, this is not usually considered problematic. A normalization technique can be used to correct for this effect during the image processing stage. However, to simulate a real-world situation where this device might be used, normalization was not carried out in this study. Note that prolonged exposure of UV or violet light can be harmful to the human eye and thus a device which can capture fluorescence emissions and display them on an external monitoring system is safer. Other benefits of using a digital image capture system are that the images can be stored for safety inspection records, the images can be observed and analysed outside the food processing environment, and the images can be viewed on multiple image display devices (Lee et al., 2013).

4. Conclusions Assessment of an HFID as an aid for detection of food residues on HDPE and SS food processing surfaces was carried out for three foods which have been associated with foodborne illness outbreaks. The HFID assessed in this study allowed the user to employ appropriate optical filters to optimize the fluorescence emission contrast between specific food residues and processing surfaces. Food residues which were difficult to identify by the human eye were easily identified using the HFID combined with basic image analysis techniques. Identification of these food residues and subsequent removal could be employed to reduce the risks of biofilm formation and cross-contamination due to inadequate cleaning and sanitization procedures in the food processing industry. Reducing cross-contamination of food products with pathogenic bacteria

would reduce incidents of foodborne illnesses and the associated costs. Acknowledgements This publication is the result of research conducted with the financial support of the European Union’s Seventh Framework Programme (FP7) under the Marie Curie International Outgoing Fellowships for Career Development (FP7-PEOPLE-2011-IOF; grant number, PIOF-GA-2011-299392). References Agle, M. E. (2007). Biofilms in the food industry. In H. P. Blaschek, H. H. Wang, & M. E. Agle (Eds.), Biofilms in the food environment (pp. 3e18). Oxford, UK: Blackwell Publishing Ltd. Aslam, M., Greer, G. G., Nattress, F. M., Gill, C. O., & McMullen, L. M. (2004). Genotypic analysis of Escherichia coli recovered from product and equipment at a beef-packing plant. Journal of Applied Microbiology, 97, 78e86. Aycicek, H., Oguz, U., & Karci, K. (2006). Comparison of results of ATP bioluminescence and traditional hygiene swabbing methods for the determination of surface cleanliness at a hospital kitchen. International Journal of Hygiene and Environmental Health, 209(2), 203e206. Barish, J. A., & Goddard, J. M. (2014). Stability of nonfouling stainless steel heat exchanger plates against commercial cleaning agents. Journal of Food Engineering, 124, 143e151. Birlouez-Aragon, I., Nicolas, M., Metais, A., Marchond, N., Grenier, J., & Calvo, D. (1998). A rapid fluorimetric method to estimate the heat treatment of liquid milk. International Dairy Journal, 8, 771e777. Birlouez-Aragon, I., Sabat, P., & Gouti, N. (2002). A new method for discriminating milk heat treatment. International Dairy Journal, 12, 59e67. Boulange-Petermann, L. (1996). Processes of bioadhesion on stainless steel surfaces and cleanability: a review with a special reference to the food industry. Biofouling, 10, 275e300. Boulange-Petermann, L., Jullien, C., Dubois, P. E., Benezech, T., & Faille, C. (2004). Influence of surface chemistry on the hygienic status of industrial stainless steel. Biofouling, 20(1), 25e33. Chmielewski, R. A. N., & Frank, J. F. (2007). Inactivation of Listeria monocytogenes biofilms using chemical sanitizers and heat. In H. P. Blaschek, H. H. Wang, & M. E. Agle (Eds.), Biofilms in the food environment (pp. 73e104). Oxford, UK: Blackwell Publishing Ltd. Cho, B. K., Chen, Y. R., & Kim, M. S. (2007). Multispectral detection of organic residues on poultry processing plant equipment based on hyperspectral reflectance imaging technique. Computers and Electronics in Agriculture, 57, 177e189. Christensen, J., Becker, E. M., & Frederiksen, C. S. (2005). Fluorescence spectroscopy and PARAFAC in the analysis of yogurt. Chemometrics and Intelligent Laboratory Systems, 75, 201e208. Cole, M. B., Jones, M. V., & Holyoak, C. (1990). The effect of pH, salt concentration and temperature on the survival and growth of Listeria monocytogenes. Journal of Applied Bacteriology, 69, 63e72. Corp, L. A., McMurtrey, J. E., Chappelle, E. W., Daughtry, C. S. T., & Kim, M. S. (1997). UV band fluorescence (in vivo) and its applications to the remote assessment of N fertilization level. Remote Sensing of Environment, 61, 110e117.

C.D. Everard et al. / Food Control 59 (2016) 243e249 Davidson, C. A., Griffith, C. J., Peters, A. C., & Fielding, L. M. (1999). Evaluation of two methods for monitoring surface cleanliness-ATP bioluminescence and traditional hygiene swabbing. Luminescence: The Journal of Biological and Chemical Luminescence, 14(1), 33e38. De Jong, P. (1997). Impact and control of fouling in milk processing. Trends in Food Science & Technology, 8(12), 401e405. Donlan, R. M., & Costerton, J. W. (2002). Biofilms: survival mechanisms of clinically relevant microorganisms. Clinical Microbiology Reviews, 15(2), 167e193. Dourou, D., Beauchamp, C. S., Yoon, Y., Geornaras, I., Belk, K. E., Smith, G. C., et al. (2011). Attachment and biofilm formation by Escherichia coli O157:H7 at different temperatures, on various food-contact surfaces encountered in beef processing. International Journal of Food Microbiology, 149, 262e268. Everard, C. D., Kim, M. S., & Lee, H. (2014). A comparison of hyperspectral reflectance and fluorescence imaging techniques for detection of contaminants on spinach leaves. Journal of Food Engineering, 143, 139e145. Gill, C. O., & McGinnis, J. C. (2000). Contamination of beef trimmings with Escherichia coli during a carcass breaking process. Food Research International, 33, 125e130. Gun, H., Yilmaz, A., Turker, S., Tanlasi, A., & Yilmaz, H. (2003). Contamination of bovine carcasses and abattoir environment by Escherichia coli O157:H7 in Istanbul. International Journal of Food Microbiology, 84, 339e344. Hadjiev, D., Dimitrov, D., Martinov, M., & Sire, O. (2007). Enhancement of the biofilms formation on polymeric supports by surface conditioning. Enzyme and Microbial Technology, 40, 840e848. Haeghebaert, S., Le Querrec, F., Vaillant, V., Delarocque Astagneau, E., & Bouvet, P. (2001). Les toxi-infections alimentaires collectives en France en 1998. Bulletin Epid emiologique Hebdomadaire, 15, 1e12. Hood, S. K., & Zottola, E. A. (1995). Biofilms in food processing: review. Food Control, 6(1), 9e18. Jessen, B., & Lammert, L. (2003). Biofilm and disinfection in meat processing plants. International Biodeterioration & Biodegradation, 51(4), 265e269. Jun, W., Kim, M. S., Cho, B. K., Millner, P. D., Chao, K., & Chan, D. E. (2010). Microbial biofilm detection on food contact surfaces by macro-scale fluorescence imaging. Journal of Food Engineering, 99, 314e322. Kim, M. S., Lefcourt, A. M., & Chen, Y. R. (2003). Optimal fluorescence excitation and emission bands for detection of fecal contamination. Journal of Food Protection, 66(7), 1198e1207. Kim, M. S., McMurtrey, J. E., Mulchi, C. L., Daughtry, C. S. T., Chappelle, E. W., & Chen, Y. R. (2001). Steady-state multispectral fluorescence imaging system for plant leaves. Applied Optics, 40(1), 157e166. Koo, O. K., Martin, E. M., Story, R., Lindsay, D., Ricke, S. C., & Crandall, P. G. (2013). Comparison of cleaning fabrics for bacterial removal from food-contact surfaces. Food Control, 30, 292e297. Koo, O. K., Ndahetuye, J. B., O’Bryan, C. A., Ricke, S. C., & Crandall, P. G. (2014). Influence of Listeria innocua on the attachment of Listeria monocytogenes to stainless steel and aluminum surfaces. Food Control, 39, 135e138. Koutsoumanis, K., & Sofos, J. N. (2004). Microbial contamination. In W. K. Jensen (Ed.), Encyclopedia of meat sciences (pp. 727e737). Amsterdam, The Netherlands: Elsevier Academic Press. Koutzayiotis, C. (1992). Bacterial biofilms in milk pipelines. African Journal of Dairy Science, 24, 19e22. Le Gentil, C., Sylla, Y., & Faille, C. (2010). Bacterial re-contamination of surfaces of food processing lines during cleaning in place procedures. Journal of Food Engineering, 96, 37e42. Lee, H., Kim, M. S., Chao, K., Lefcourt, A. M., & Chan, D. E. (2013). Development of fluorescence based handheld imaging devices for food safety inspection. In Proceedings SPIE 8721, sensing for agriculture and food quality and safety V, 872109. vre, C., Antonini, G., Faille, C., & Be ne zech, T. (2002). Cleaning-in-place: Lelie modelling of cleaning kinetics of pipes soiled by Bacillus spores assuming a process combining removal and deposition. Food and Bioproducts, 80, 305e312. Liu, X., & Metzger, L. E. (2007). Application of fluorescence spectroscopy for monitoring changes in nonfat dry milk during storage. Journal of Dairy Science, 90, 24e37.

249

Martinon, A., Cronin, U. P., Quealy, J., Stapleton, A., & Wilkinson, M. G. (2012). Swab sample preparation and viable real-time PCR methodologies for the recovery of Escherichia coli, Staphylococcus aureus or Listeria monocytogenes from artificially contaminated food processing surfaces. Food Control, 24, 86e94. Moore, G., & Griffith, C. (2002). A comparison of surface sampling methods for detecting coliforms on food contact surfaces. Food Microbiology, 19, 65e73. Nocker, A., Cheung, C. Y., & Camper, A. K. (2006). Comparison of propidium monoazide with ethidium monoazide for differentiation of live vs. dead bacteria by selective removal of DNA from dead cells. Journal of Microbiological Methods, 67(2), 310e320. O’Toole, G. A., & Kaplan, H. B. (2000). Biofilm formation as microbial development. Annual Reviews in Microbiology, 49e79. Painter, J. A., Hoekstra, R. M., Ayers, T., Tauxe, R. V., Braden, C. R., Angulo, F. J., et al. (2013). Attribution of foodborne illnesses, hospitalizations, and deaths to food commodities by using outbreak data, United States, 1998e2008. Emerging Infectious Diseases, 19(3). Palmer, J., Flint, S., & Brooks, J. (2007). Bacterial cell attachment, the beginning of a biofilm. Journal of Industrial Microbiology and Biotechnology, 34, 577e588. Parkar, S. G., Flint, S. H., & Brooks, J. D. (2004). Evaluation of the effect of cleaning regimes on biofilms of thermophilic bacilli on stainless steel. Journal of Applied Microbiology, 96(1), 110e116. rez-Rodríguez, F., Valero, A., Carrasco, E., García, R. M., & Zurera, G. (2008). UnPe derstanding and modelling bacterial transfer to foods: a review. Trends in Food Science & Technology, 19(3), 131e144. Pompermayer, D. M. C., & Gaylarde, C. C. (2000). The influence of temperature on the adhesion of mixed cultures of Staphylococcus aureus and Escherichia coli to polypropylene. Food Microbiology, 17, 361e365. Ramanujam, N. (2000). Fluorescence spectroscopy in vivo. In R. A. Meyers (Ed.), Encyclopedia of analytical chemistry (pp. 20e56). Chichester, UK: John Wiley & Sons Ltd. Reij, M. W., & Den Aantrekker, E. D. (2004). ILSI Europe risk analysis in microbiology task force, recontamination as a source of pathogens in processed foods. International Journal of Food Microbiology, 91, 1e11. Salo, S., Ehavald, H., Raaska, L., Vokk, R., & Wirtanen, G. (2006). Microbial surveys in Estonian dairies. LWT e Food Science and Technology, 39(5), 460e471. Sharma, M., & Anand, S. K. (2002). Characterization of constitutive microflora of biofilms in dairy processing lines. Food Microbiology, 19, 627e636. Silagyi, K., Kim, S. H., Lo, Y. M., & Wei, C. I. (2009). Production of biofilm and quorum sensing by Escherichia coli O157:H7 and its transfer from contact surfaces to meat, poultry, ready-to-eat deli, and produce products. Food Microbiology, 26(5), 514e519. ~ es, M., Simo ~es, L. C., Machado, I., Pereira, M. O., & Vieira, M. J. (2006). Control of Simo flow-generated biofilms with surfactants: evidence of resistance and recovery. Food and Bioproducts, 84(4), 338e345. Srey, S., Jahid, I. K., & Ha, S. D. (2013). Biofilm formation in food industries: a food safety concern: review. Food Control, 31, 572e585. Waak, E., Tham, W., & Danielsson-Tham, M. L. (2002). Prevalence and fingerprinting of Listeria monocytogenes strains isolated from raw whole milk in farm bulk tanks and in dairy plant receiving tanks. Applied and Environmental Microbiology, 68(7), 3366e3370. Whipps, J. M., Hand, P., Pink, D. A. C., & Bending, G. D. (2008). Chapter 7 human pathogens and the phyllosphere. Advances in Applied Microbiology, 64, 183e221. Wiederoder, M. S., Liu, N., Lefcourt, A. M., Kim, M. S., & Lo, Y. M. (2013). Use of a portable hyperspectral imaging system for monitoring the efficacy of sanitation procedures in produce processing plants. Journal of Food Engineering, 117, 217e226. Wold, J. P., & Kvaal, K. (2000). Mapping lipid oxidation in chicken meat by multispectral imaging of auto fluorescence. Applied Spectroscopy, 54, 900e909. Wold, J. P., Lundby, F., & Egelandsdal, B. (1999). Quantification of connective tissue (hydroxyproline) in ground beef by autofluorescence spectroscopy. Journal of Food Science, 64, 377e383.