Augmented Reality for Real-Time Detection and Interpretation of

Subscriber access provided by Binghamton University | Libraries

Article

Augmented Reality for Real-time Detection and Interpretation of Colorimetric Signals Generated by Paper-Based Biosensors Steven Russell, Antonio Domenech-Sanchez, and Roberto de la Rica ACS Sens., Just Accepted Manuscript • Publication Date (Web): 26 May 2017 Downloaded from http://pubs.acs.org on May 27, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Sensors is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

Augmented Reality for Real-time Detection and Interpretation of Colorimetric Signals Generated by Paper-Based Biosensors

Steven M. Russell,1 Antonio Doménech-Sánchez,2,3 Roberto de la Rica.1,*

1

Department of Chemistry, University of the Balearic Islands, Carretera de Valldemossa km 7.5, 07122 Palma de Mallorca, Illes Balears, Spain. 2

3

Saniconsult Ibérica SL, Palma de Mallorca, España

Área de Microbiología e Instituto Universitario de Investigación en Ciencias de la Salud (IUNICS), University of the Balearic Islands, Palma de Mallorca, Spain.

Corresponding author: *[email protected]

ACS Paragon Plus Environment

1

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 22

Abstract. Colorimetric tests are becoming increasingly popular in point-of-need analyses due to the possibility of detecting the signal with the naked eye, which eliminates the utilization of bulky and costly instruments only available in laboratories. However, colorimetric tests may be interpreted incorrectly by non-specialists due to disparities in color perception or a lack of training. Here we solve this issue with a method that not only detects colorimetric signals but also interprets them so that the test outcome is understandable for anyone. It consists of an augmented reality (AR) app that uses a camera to detect the colored signals generated by a nanoparticle-based immunoassay, and that yields a warning symbol or message when the concentration of analyte is higher than a certain threshold. The proposed method detected the model analyte mouse IgG with a limit of detection of 0.3 µg mL-1, which was comparable to the limit of detection afforded by classical densitometry performed with a non-portable device. When adapted to the detection of E. coli, the app always yielded a “hazard” warning symbol when the concentration of E. coli in the sample was above the infective dose (106 cfu mL-1 or higher). The proposed method could help non-specialists make a decision about drinking from a potentially contaminated water source by yielding an unambiguous message that is easily understood by anyone. The widespread availability of smartphones along with the inexpensive paper test that requires no enzymes to generate the signal makes the proposed assay promising for analyses in remote locations and developing countries.

Keywords: Gold, Nanoparticles, Immunosensor, Pathogens, Bacteria, Smartphone, Water, Augmented Reality


2

Page 3 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

Colorimetric biosensors are becoming increasingly popular for the detection of analytes at the point of need when the signal generated by the target molecule can be detected with the naked eye.1 This visual detection strategy means that no laboratory equipment is required to read the signal, which reduces costs and enhances the portability for infield measurements. Low-cost and highly portable detection systems are particularly useful in developing countries where instrument-based detection methods are less likely to be available or accessible to those who need it most. 2 However naked-eye detection also has limitations. For example, individual differences in vision and color perception may pose problems for detecting the signal clearly, even with a chart of analyte concentrations and their corresponding colors.3 This is particularly problematic when the test is performed by non-specialists, who may lack the training to interpret the test results with confidence.4

A potential solution to these problems is to develop apps that use the camera of a mobile device as a transducer to detect colors on a substrate.5 Mobile devices are portable and readily available worldwide, and therefore they can easily be implemented at the point of need without increasing costs or sacrificing the portability of the assay. However, these apps quantify color levels in photographs taken with a camera, and therefore the outcome of the test is highly dependent on ambient light conditions as well as on the distance and angle between the camera and the sample.6–10 This may result in false positive or false negative results when the detection is not performed under lightcontrolled conditions, or when the camera is not placed in the exact same position. Furthermore, the outcome of the test is usually an array of numbers, which, as before, may be ambiguous and difficult to interpret by non-specialists. To overcome these


3

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 22

obstacles, it would be desirable to find a method that was able to not only read the colorimetric signal in a wide range of ambient light conditions, but also to interpret the results of the test in a way that was easily understood by the user. Such a method could help non-specialists make an informed decision rapidly and at the point of need.

In this article we propose using the target recognition capabilities of augmented reality (AR) to detect colorimetric signals in real time. The colorimetric signal consists of colored spots generated on paper substrates by gold nanoparticle-labeled antibodies (Figure 1a).11 AR has been previously utilized to create digital media for entertainment, advertising, and medical devices.12 Our AR app has been developed to generate digitally augmented messages upon recognition of a unique complex pattern, in this case a QR code (Fig. 1b). The QR code is printed on a transparent sheet, which is superimposed on the test. When the analyte is absent, or found at low concentrations, the colorimetric signal is weak, and therefore does not disrupt the QR code pattern. Under this condition the QR code is recognized and the AR app generates a symbol indicating that the test is negative (Fig. 1b). When the analyte is found at high concentrations it generates intensely colored spots that disrupt the QR code pattern. Under this condition the AR app does not recognize the target image and the test result is positive. A control QR code, which should always be recognized by the app, is also added to rule out any artifacts originating from colored samples or extreme light conditions. Since the signal is generated via pattern recognition rather than through the quantification of color in an image, the outcome of the test is less affected by ambient light conditions as compared to previous approaches based on the densitometric analysis of colored spots. This is because in the latter, the pixel density in a colored spot changes as a function of the background illuminance (Fig. S3). It is also less dependent on the


4

Page 5 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

relative position and angle between the camera and the substrate, since the pattern recognition software also includes object tracking, which allows it to recognize target images from a large range of angles and distances (see Figure S7 in the Supporting Information). Furthermore, the AR software can be easily programmed to yield test outcomes in the form of warning symbols, messages or videos that can be easily interpreted by non-specialists and that can help them make an informed decision depending on the outcome of the test. Here we prove this concept by developing biosensors that yield a “hazard” or “no hazard” message when E. coli is present at concentrations higher than the infective dose threshold, and propose that they could help the user decide to not drink contaminated water, which in turn could prevent contracting infectious diseases (Fig. 1c). The easy interpretation of the assay along with the robust biosensing approach, which yields results within 30 min and does not require labile enzymes to generate the colorimetric signal, make our method promising for testing the quality of drinking water at the point of need.

Materials and Methods

Materials. WhatmanTM 41 ashless quantitative filter paper was used as substrate for the colorimetric tests. Immunoglobulin from mouse serum (mouse IgG, technical grade, >95%), anti-mouse IgG (whole molecule)-biotin antibody produced in goat (biotin antimouse IgG, affinity isolated antibody), poly(ethylene glycol) 2-mercaptoethyl ether acetic

acid

(SH-PEG-COOH,

Mn

2100),

N-(3-Dimethylaminopropyl)-N′-

ethylcarbodiimide hydrochloride (EDC, purum >98%), N-hydroxysulfosuccinimide sodium salt (sulfo-NHS, >98%), MES monohydrate (>98%), Gold(III) chloride hydrate (99.999%), sodium citrate tribasic dihydrate (>98%), Tween-20 and bovine serum


5

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 22

albumin (BSA, >98%) were purchased from Sigma-Aldrich. Biotinylated anti-E. coli serotype O/K polyclonal antibody produced in goat (anti-E. coli) was purchased from ThermoFischer. Avidin from egg white was purchased from Calbiochem. Phosphate buffer saline (PBS, pH 7.4) was prepared following standard procedures available elsewhere. PBST refers to PBS containing 0.1% Tween-20. PBS-BSA refers to a PBS solution containing 5 mg mL-1 BSA. Escherichia coli ATCC® 25922 was purchased from Microkit, and grown and diluted in Tryptic Soy Broth (TSB, Sharlau) to a final concentration of 109 cfu mL-1. Spiked samples were prepared by diluting this sample in uperized whole milk that was used immediately after opening the sterile container. E. coli cells were previously killed by heating the sample at 95oC for 10 minutes in order to avoid their growth during the experiments.

Synthesis and modification of gold nanoparticles. Citrate-capped gold nanoparticles with LSPR centered at 530 nm and an average diameter of 40 nm were obtained with the Turkevich method (Figure S1). Briefly, 29 mg of sodium citrate were added to 28 mg of gold chloride in 250 mL of boiling water under agitation and let react for 15 min. The nanoparticles were modified with 0.1 mM SH-PEG-COOH overnight. Then the nanoparticles were concentrated 50 times by centrifugation (10 min, 13,000 rpm) and washed 3 times with water. The resulting pegylated nanoparticles were covalently modified with avidin with the following procedure. First, the nanoparticles were suspended in 0.5 M MES buffer pH 5.5 by centrifugation. EDC and sulfo-NHS were then added to a final concentration of 10 mM. After 30 min the nanoparticles were centrifuged, the solution was removed and 1 mL of avidin (1 mg mL-1 in PBS) was added. The solution was left overnight at 4oC. The next day, 100 µL of 0.1 M glycine was added for 30 min to cap any unreacted NHS esters followed by washing three times


6

Page 7 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

with PBS. The nanoparticles were kept at 4oC in PBST supplemented with 0.05% sodium azide.

Detection of mouse IgG and E. coli with nanoparticle probes. 1 mg mL-1 mouse IgG or 109 cfu mL-1 E. coli was serially diluted with PBS to different final concentrations. 2 µL of the resulting samples was spotted 5 times on a piece of filter paper in order to generate tests that could block the recognition of QR codes (see Section S2 in the Supporting Information). After 5 min, the paper substrates were immersed in PBS-BSA for 5 min. Subsequently the tests were placed on a clean piece of paper and 2 µL of biotinylated antibody was applied to each spot. Biotinylated anti-mouse IgG diluted 1:10 in PBST was used for detecting mouse IgG whereas biotinylated anti-E. coli diluted 1:50 in PBST was utilized to detect E. coli specifically. The paper substrates were incubated for 15 min in a humid chamber followed by a washing step in PBST (30 s). Then 1 µL of avidin-modified gold nanoparticles diluted 1:2 in PBST was spotted for 4 min followed by a 1 min washing step in PBST.

Colorimetric signal evaluation with an AR app. Results in Figures 3a and 4 were obtained as follows. QR codes were printed on high resistance polyester films from APLI with a HP Envy printer using black ink from HP. Details about QR code design and assay optimization can be found in the Supporting Information (Section S2). It should be noted that our QR codes do not direct the user to a website as previously proposed in other detection systems.13 QR codes are used here because they can be easily obtained with a wide variety of unique patterns, and therefore they are ideal targets for designing AR apps. The sampling pattern was superimposed on the colorimetric test whereas the control pattern was superimposed on a piece of paper


7

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 22

incubated with the matrix and subject to the same blocking and wash steps. This was done to evaluate any potential artifacts arising from the color of the sample or the wetting of the paper substrate. Disruption of the QR code pattern by the underlying colored spots was evaluated with an AR app developed using Vuforia (version 6.2) through a digital camera (Havit HD web camera containing a HD Color OV 0308 CMOS Sensor and a 4P 2M Lends F2.8 lens). A positive result triggered the apparition of a symbol or word within a few seconds only on the control QR code, whereas a negative result resulted in the generation of a symbol or word on both sampling and control QR codes (Figures 1b-1c). The tests were considered invalid if the control did not generate a symbol or word. Three independent tests were performed to evaluate the variability of the assay. Videos showing the outcome of each experiment were recorded. Sample videos showing a positive and negative test outcome for each analyte can be found in the Supporting Information. Other videos are available by request from the corresponding author. Snapshots showing the results of all the experiments are available as Figures S9 and S10 in the Supporting Information.

The proposed detection system was compared to a classical densitometry approach. To this end, the tests were imaged with a flatbed scanner and transformed into black and white images with Adobe Photoshop. The color was then inverted and the average pixel density (luminosity) inside each spot was obtained with the histogram function of Photoshop. The densitometric signal of each assay was taken as the average value of the 5 spots. The final densitometric signal displayed in Figure 3b is the average of the 3 independent assays. Error bars are the standard deviation.


8

Page 9 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

ELISA for mouse IgG. The proposed nanoparticle-based assays were compared to a classical laboratory-based enzyme-linked immunosorbent assay (ELISA) performed as follows. 100 µL of mouse IgG in PBS at different final concentrations was added to a Nunc Maxisorp 96-well plate for 1 h followed by blocking with PBS-BSA for 1 h and washing once with PBST. Then biotinylated anti-mouse IgG diluted 1:300 in PBST was added for 15 min followed by washing 3 times with PBST and incubating with streptavidin-HRP diluted 1:100 in PBST for 30 min. The plates were then washed 5 times with PBST and 100 µL of 0.1 mg mL-1 3,3′,5,5′-tetramethylbenzidine (TMB) in 50 mM acetate buffer pH 5 supplemented with 0.03% H2O2 was added to each well. The enzymatic reaction was stopped by adding 100 µL of 2 M sulfuric acid to each well. The absorbance at 450 nm of the resulting yellow-colored solutions was measured with a Sunrise plate reader.

Results and Discussion

The effect of different ambient light conditions in the target recognition capabilities of the proposed AR app was tested as follows. The AR app was programmed to produce a cube upon recognition of the control pattern and a sphere when recognizing the sampling pattern. The target patterns (QR codes) were then superimposed onto a white piece of paper and imaged with a camera under different illuminance conditions, which were measured with a lux meter. In Figure 2 the AR app recognizes the patterns and yields the same test outcome when the illuminance is between 100 lux (very dark overcast day) and 500 lux (well lit office). However, the densitometric analysis of the same images yields different results for each light condition because the pixel density (color brightness) in the target image changes with the background illuminance (Figure


9

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 22

S6). These results are in line with our initial hypothesis that the proposed detection method is more robust towards changes in ambient light conditions compared to traditional approaches using densitometry. This means that there is no need to create a lightbox to control the illuminance, nor is it required to apply complex algorithms to compensate for ambient light effects in the detection of the colorimetric signals.8,10,14

After studying the effect of ambient light in signal transduction we prepared tests for the detection of a model analyte (mouse IgG) following the procedure detailed in Figure 1a. The test outcomes were detected with the proposed method as well as with densitometry. The latter was accomplished by imaging the tests with a flatbed scanner. The flatbed scanner was used instead of a camera in order to avoid changes in ambient light conditions leading to differences in color quantification (Figure S6). Furthermore, the proposed tests were also compared to a laboratory-based ELISA, which lasted several hours. Results are summarized in Figure 3. Figure 3a shows the test outcome (positive or negative) as a function of the concentration of mouse IgG obtained with the AR app (see also Figure S8). The three independent calibration experiments yielded exactly the same results. The limit of detection, defined as the concentration of analyte that yields a positive result in three independent experiments, was 0.3 µg mL-1. The assays always yielded a negative outcome when the concentration of analyte was lower than this value. When the same tests were imaged with a flatbed scanner and quantified with densitometry, a limit of detection of 0.3 µg mL-1 based on the 2σ criterion was also obtained. In Figure 3c using enzymes to amplify the colorimetric signals decreased the limit of detection to 0.01 µg mL-1.


10

Page 11 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

In light of the results obtained in Figure 3 the three methods can be compared as follows. ELISA yields quantitative results with the lowest limit of detection, as expected from an enzyme-amplified approach. The signal is linear between 0.01 and 0.3 µg mL-1 (y = 0.52log(x) + 1.2, r2 = 0.9997). However, the longsome laboratory-based procedure and instrumentation required for this method along with the labile nature of the enzyme, which needs to be constantly refrigerated, make this approach unsuitable for analyses at the point of need. Densitometry also yields quantitative results that enable determining the concentration of the analyte with a dynamic range expanding from 0.3 to 10 µg mL-1 (y = 36log(x) + 38, r2 = 0.998). Yet, it should be noted that the results shown in Figure 3 were obtained with a non-portable piece of equipment (a flatbed scanner). Indeed, when the densitometric analysis was performed using pictures taken with the camera of a cell phone (photographs shown in Fig. 3a), the limit of detection increased to 1 µg mL-1 (see Fig. S7). Compared to these methods the proposed AR app is more advantageous in that it can yield highly trustable qualitative data immediately and using a portable mobile device. The tests could be adapted to yield a positive signal above a different threshold value by using darker or brighter QR codes. Brighter QR codes allow for detecting less intense colorimetric signals, and therefore for more sensitive detections. For example, in Figure S5 (Supporting Information) when the brightness of the QR code is 67 (HSB palette) the AR software recognizes colored spots with brightness 88 or lower, whereas a darker QR code with a brightness of 63 only allows for the detection of darker colored patterns with brightness 78 or lower. However, brighter QR codes may suffer more from interferences originating from nonspecific interactions or changes in the color of the paper matrix. The choice of a particular QR code, and therefore the concentration at which the assay outcome switches from negative to positive, is a balance between the sensitivity required to


11

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 22

detect a particular analyte and the need to eliminate interferences. With our approach, the person performing the analysis only needs to position a camera on top of the test for the app to generate a result in real time. While this method does not enable quantifying the signal, it can be adapted to provide a piece of information that can be easily interpreted by non-specialists. Below we explore this concept for producing tests that advise the user on the danger of drinking water potentially contaminated with bacteria. According to the World Health Organization contaminated drinking-water is estimated to cause more than 0.5 million diarrheal deaths each year. These deaths could be prevented if a simple test was available to non-specialists in order to detect hazardous levels of bacteria at the point of need. With this in mind we adapted the nanoparticlebased immunoassay in Fig. 1a for the detection of E. coli and programmed the AR software to yield an easy to interpret “hazard” or “no hazard” test outcome depending on the concentration of analyte (Fig. 1c) (see also Figure S9). The test consisted of letting a drop of sample dry on the paper, which physically adsorbed the analyte on the substrate. The the target pathogen was then specifically detected using a biotinylated antibody against E. coli. This procedure is advantageous in that it only requires one antibody to detect the analyte, which reduces costs compared to sandwich-type immunoassay formats. The three independent calibration experiments yielded the exact same results. In Figure 4 the “hazard” word appears in all the assays when the concentration of E. coli is higher than 106 cfu mL-1, which is the lowest infective dose recommended for the detection of pathogenic strains of these bacteria.15 This means that the assay never yielded a false negative above the infective dose, and therefore that the user was never advised to drink hazardous water. The proposed detection method also discriminates false negatives and yields a “no hazard” symbol when the concentration of bacteria is lower than the infective dose in all three independent experiments. To test


12

Page 13 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

whether the assay could be used to detect bacteria in a more complex matrix, E. coli were spiked into whole milk and the assay was repeated in the same conditions. In Figure 4b the assay yields a “hazard” warning symbol when the concentration of pathogens in the milk is equal or higher than 106 cfu mL-1, which demonstrates that the assay is specific for E. coli and that the limit of detection is not affected by the matrix. While previous approaches for detecting bacteria yielded colors or numbers that needed to be interpreted by referencing an accompanying table,14,16–18 our app yields straightforward warning symbols containing relevant information for the user. Such an approach could be particularly useful when the test is performed by unskilled users of limited educational background.

Conclusions In conclusion we have demonstrated that the pattern recognition ability of an AR app can be used to detect colorimetric signals with the same limit of detection as a densitometry-based approach that was performed with a non-portable scanner under controlled light conditions. The AR app can be programmed to concomitantly interpret the signal by yielding warning symbols or advice messages when the concentration of analyte is higher than a certain threshold. This can help non-specialists make an informed decision rapidly after performing the assay. For example, the AR app programmed to yield a “hazard” outcome when the concentration of E. coli is higher than the infective dose could help users make the decision to not drink contaminated water or milk, which in turn could prevent them from contracting infectious diseases. The proposed nanoparticle-based immunoassay also presents several advantages for infield detection including the implementation of low-cost and disposable paper as substrate,19 the utilization of sturdy nanoparticles rather than labile reagents to generate


13

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 22

the signals, and a short assay time within 30 min. Further work is under development in order to fully automatize all the steps of the immunoassay for the easy utilization of the test at the point of need. The method only requires printing a QR code on an inexpensive transparency and positioning a mobile device over it. Precise control over the position of the device is not needed, nor are signal correction algorithms required, which expedites the detection process. Moreover, the proposed detection method could be adapted for reading and interpreting the signals generated by other commonly-used colorimetric biosensors such as lateral flow tests

20

and ELISA21,22 when the colored

patterns or solutions generated by the test disrupt a sampling pattern positioned onto them. Using lateral flow tests would reduce the number of analytical steps and time required for the detection, which is also advantageous in decentralized studies. The sensitivity of these approaches could be boosted by using dual chemiluminescent-gold probes23 or with nucleic acid amplification strategies.24

Supporting Information Available: The following files are available free of charge. Supporting Information.pdf: Contains Figures S1 to S10 VideoS1.avi: negative result in detection of mouse IgG. VideoS2.avi: positive result in the detection of mouse IgG. VideoS3.avi: negative result in the detection of E. coli. VideoS4.avi: negative result in the detection of E. coli.

Acknowledgements Dr. de la Rica acknowledges a Ramón y Cajal contract from MICINN.

The authors declare no competing financial interest.


14

Page 15 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

References (1)

Paterson, S.; de la Rica, R. Solution-Based Nanosensors for in-Field Detection with the Naked Eye. Analyst 2015, 140, 3308–3317.

(2)

Boppart, S. A.; Richards-Kortum, R. Point-of-Care and Point-of-Procedure Optical Imaging Technologies for Primary Care and Global Health. Sci. Transl. Med. 2014, 6, 253rv2-253rv2.

(3)

Kumar, A. a.; Hennek, J. W.; Smith, B. S.; Kumar, S.; Beattie, P.; Jain, S.; Rolland, J. P.; Stossel, T. P.; Chunda-Liyoka, C.; Whitesides, G. M. From the Bench to the Field in Low-Cost Diagnostics: Two Case Studies. Angew. Chemie Int. Ed. 2015, 54, 5836–5853.

(4)

Kalluri, P.; Naheed, A.; Rahman, S.; Ansaruzzaman, M.; Faruque, A. S. G.; Bird, M.; Khatun, F.; Bhuiyan, N. A.; Nato, F.; Fournier, J. M.; et al. Evaluation of Three Rapid Diagnostic Tests for Cholera: Does the Skill Level of the Technician Matter? Trop. Med. Int. Heal. 2006, 11, 49–55.

(5)

Roda, A.; Michelini, E.; Zangheri, M.; Di Fusco, M.; Calabria, D.; Simoni, P. Smartphone-Based Biosensors: A Critical Review and Perspectives. TrAC Trends Anal. Chem. 2016, 79, 317–325.

(6)

Wang, L. J.; Chang, Y. C.; Sun, R.; Li, L. A Multichannel Smartphone Optical Biosensor for High-Throughput Point-of-Care Diagnostics. Biosens. Bioelectron. 2017, 87, 686–692.

(7)

Chun, H. J.; Park, Y. M.; Han, Y. D.; Jang, Y. H.; Yoon, H. C. Paper-Based Glucose Biosensing System Utilizing a Smartphone as a Signal Reader. Biochip J. 2014, 8, 218–226.

(8)

Vashist, S. K.; van Oordt, T.; Schneider, E. M.; Zengerle, R.; von Stetten, F.;


15

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 22

Luong, J. H. T. A Smartphone-Based Colorimetric Reader for Bioanalytical Applications Using the Screen-Based Bottom Illumination Provided by Gadgets. Biosens. Bioelectron. 2015, 67, 248–255. (9)

Yetisen, A. K.; Martinez-Hurtado, J. L.; Garcia-Melendrez, A.; Da Cruz Vasconcellos, F.; Lowe, C. R. A Smartphone Algorithm with Inter-Phone Repeatability for the Analysis of Colorimetric Tests. Sensors Actuators, B Chem. 2014, 196, 156–160.

(10)

Masawat, P.; Harfield, A.; Namwong, A. An iPhone-Based Digital Image Colorimeter for Detecting Tetracycline in Milk. Food Chem. 2015, 184, 23–29.

(11)

Parolo, C.; Merkoçi, A. Paper-Based Nanobiosensors for Diagnostics. Chem. Soc. Rev. 2013, 42, 450–457.

(12)

Khor, W. S.; Baker, B.; Amin, K.; Chan, A.; Patel, K.; Wong, J. Augmented and Virtual Reality in Surgery—the Digital Surgical Environment: Applications, Limitations and Legal Pitfalls. Ann. Transl. Med. 2016, 4, 454–454.

(13)

Yuan, M.; Liu, K. K.; Singamaneni, S.; Chakrabartty, S. Self-Powered Forward Error-Correcting Biosensor Based on Integration of Paper-Based Microfluidics and Self-Assembled Quick Response Codes. IEEE Trans. Biomed. Circuits Syst. 2016, PP, 963–971.

(14)

Adkins, J. A.; Boehle, K.; Friend, C.; Chamberlain, B.; Bisha, B.; Henry, C. S. Colorimetric and Electrochemical Bacteria Detection Using Printed Paper- and Transparency-Based Analytic Devices. Anal. Chem. 2017, 89, 3613−3621.

(15)

Kothary, M. H.; Babu, U. S. Infective Dose of Foodborne Pathogens in Volunteers: A Review. J. Food Saf. 2001, 21, 49–68.

(16)

De La Rica, R.; Pejoux, C.; Fernandez-Sanchez, C.; Baldi, A.; Matsui, H. Peptide-Nanotube Biochips for Label-Free Detection of Multiple Pathogens.


16

Page 17 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

Small 2010, 6, 1092–1095. (17)

Chen, J.; Jackson, A. A.; Rotello, V. M.; Nugen, S. R. Colorimetric Detection of Escherichia Coli Based on the Enzyme-Induced Metallization of Gold Nanorods. Small 2016, 12, 2469–2475.

(18)

Joo, J.; Kwon, D.; Shin, H. H.; Park, K.-H.; Cha, H. J.; Jeon, S. A Facile and Sensitive Method for Detecting Pathogenic Bacteria Using Personal Glucose Meters. Sensors Actuators B Chem. 2013, 188, 1250–1254.

(19)

Hu, J.; Wang, S.; Wang, L.; Li, F.; Pingguan-Murphy, B.; Lu, T. J.; Xu, F. Advances in Paper-Based Point-of-Care Diagnostics. Biosens. Bioelectron. 2014, 54, 585–597.

(20) Choi, J. R.; Hu, J.; Tang, R. H.; Gong, Y.; Feng, S.; Ren, H.; Wen, T.; Li, X. (James); Wan Abas, W. A. B.; Pingguan-Murphy, B.; et al. An Integrated PaperBased Sample-to-Answer Biosensor for Nucleic Acid Testing at the Point of Care. Lab Chip 2016, 16, 611–621. (21)

Connor, E. F. O.; Paterson, S.; Rica, R. de la. Naked-Eye Detection as a Universal Approach to Lower the Limit of Detection of Enzyme-Linked Immunoassays. Anal. Bioanal. Chem. 2016, 408, 3389–3393.

(22)

Ambrosi, A.; Airò, F.; Merkoçi, A. Enhanced Gold Nanoparticle Based ELISA for a Breast Cancer Biomarker. Anal. Chem. 2010, 82, 1151–1156.

(23)

Chen, Y.; Sun, J.; Xianyu, Y.; Yin, B.; Niu, Y.; Wang, S.; Cao, F.; Zhang, X.; Wang, Y.; Jiang, X. A Dual-Readout Chemiluminescent-Gold Lateral Flow Test for Multiplex and Ultrasensitive Detection of Disease Biomarkers in Real Samples. Nanoscale 2016, 8, 15205–15212.

(24)

Jauset-Rubio, M.; Svobodová, M.; Mairal, T.; McNeil, C.; Keegan, N.; ElShahawi, M. S.; Bashammakh, A. S.; Alyoubi, A. O.; O’Sullivan, C. K. Aptamer


17

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 22

Lateral Flow Assays for Ultrasensitive Detection of β-Conglutin Combining Recombinase Polymerase Amplification and Tailed Primers. Anal. Chem. 2016, 88, 10701–10709.

Figure 1. Schematic representation of the nanoparticle-based immunoassay on paper substrates and the proposed method for detecting and interpreting the resulting colorimetric signals with the AR app; a) Detection of analytes (A) with biotinylated antibodies and avidin-decorated gold nanoparticles; b) Signal detection: a transparency containing two QR codes is superimposed on the tests and visualized through a camera,


18

Page 19 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

when the colorimetric signal is high the sampling QR code is not recognized by the AR app and the test is positive; c) Test interpretation: the AR app can be programmed to display a warning symbol or message that can be easily interpreted by non-specialist, for example a “hazard” or “no hazard” message depending on the concentration of bacteria in the sample.

Figure 2. Pattern recognition under different light conditions; a) 12.8, b) 100, c) 200, d) 300, e) 400, and f) 500 lux.


19

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 22

Figure 3. Detection of mouse IgG with a nanoparticle-based immunoassay (a, b) and ELISA (c); a) Photographs of tests results and signal detection with the proposed AR app; the three independent assays yielded the same positive or negative outcome when read with the app; b) Scanned images and densitometric analyses of the same tests; c) Photograph of ELISA plate and signal quantification with a plate reader. All scales are semi-logarithmic; error bars are the standard deviation (n=3).


20

Page 21 of 22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Sensors

Figure 4. Detection of E. coli with nanoparticle-based immunoassays and the proposed AR app; a) “hazard” results as a function of the concentration of E. coli in colony forming units (cfu) (semi-logarithmic scale); three independent assays were performed which yielded the exact same results (see also Figure S6); b) Snapshots showing the result of the same assays performed with E. coli spiked into whole milk with the concentration of 0, 104, 105, 106, 107 and 108 cfu mL-1 (from top to bottom, respectively).


21

ACS Sensors

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 22

For TOC only:


22

Augmented Reality for Real-Time Detection and Interpretation of

Recommend Documents