Urine Test Strip Quantitative Assay with a Smartphone Camera

Urine test strips for urinalysis are a common diagnostic tool with minimal costs and are used in various situations including homecare and hospitalization. The coloration scaled by the naked eye is simple, but it is suitable for semiquantitative analysis only. In this paper, a colorimetric assay is developed based on a smartphone digital camera and urine test strips. Assays of pH, albumin, glucose, and lipase activity were performed as a tool for the diagnosis of aciduria, alkaluria, glycosuria, proteinuria, and leukocyturia. The RGB color channels were analyzed in the colorimetric assay, and the assay exerted good sensitivity, and all the particular diagnoses proved to be reliable. The limits of detection for glucose (0.11 mmol/L), albumin (0.15 g/L), and lipase (2.50 U/μL) were low enough to cover the expected physiological concentration, and the range for pH was also satisfactory. The urine test strips with a camera as an output detector proved applicability to spiked urine samples, and the results were also well in comparison to the standard assays which confirms the practical relevance of the presented findings.


Introduction
Urine test strips are a common tool in clinical biochemistry that serves to diagnose various diseases, including renal disease, liver disease, and some metabolic disorders.Te analysis of urine for diagnostical purposes is frequently called a urinalysis.Proteins in urine, erythrocytes, specifc gravity (urine osmolality), nitrites, leukocyte esterase, glucose, ketones, bilirubin, and pH are the typical markers covered by standard commercial sets.Many relevant articles have widely discussed the role and importance of urinalysis for diagnostic purposes [1][2][3][4][5].
Te results achieved using urine test strips cannot fully replace the blood analysis.Some markers analyzed in the urine have less diagnostic importance than the same or similar markers in the blood.Glucose can be mentioned as an example.While the glucose level in the blood (glycemia) gradually increases, the glucose in the urine (glycosuria) occurs only after exceeding the renal glucose threshold.Tough urinalysis has some limitations compared to the analysis of blood, it also has some signifcant advantages.In addition to urinary catheterization, urine collection is a noninvasive process that can be easily performed by the patient itself or by any healthcare provider.Standard urine test strips are also quite inexpensive; they can be used without any laboratory equipment or elaborative sample pretreatment, and they can be performed in the same way as point-of-care tests.Typical urine test strips are manufactured as qualitative or semiquantitative colorimetric sensors; however, the exact determination of marker concentration is not possible without specialized equipment.
Small digital cameras integrated into common electronic devices, such as smartphones and wearable technologies, are not primarily intended as a tool for laboratory analysis.However, analytical applications based on color density channel measurement and other principles have gained popularity, and many applications have been established [6][7][8][9].Te use of smartphone cameras appears to be a promising idea in the analyses [10][11][12][13][14]. Tis paper focuses on the development of a colorimetric urinalysis based on standard commercial test strips with quantifcation of coloration by a smartphone camera.Tis approach represents a novel way in analytical chemistry to improve the standard colorimetric tests designed not for instrumental analysis but for scaling by the naked eye.Te urinalysis was purposely chosen as a test with a practical impact.It is expected that the use of the standard urinalysis tests in combination with a smartphone camera will provide accurate and more reproducible results suitable for practical use, making the assay more competitive to the standard laboratory methods but still useable, as a point-of-care test.

Materials and Methods
2.1.3D-Printed Holder.Te 3D-printed holder was made from black polyethylene terephthalate glycol with a 100 mm height and an internal tube diameter of 40 mm.Prusa Mini+ (Prusa Research; Prague, Czech Republic) printer was used.Te printer setting was the following: string diameter of 1.75 mm, infll of s50%, printing layer height of 0.1 mm, nozzle temperature 230 °C, and bed temperature of 90 °C.Te holder is shown in Figure 1.

Reagents.
Glucose, o-phenylenediamine, glucose oxidase, human serum albumin, and lipase from porcine pancreas type VI-S were purchased from Sigma-Aldrich (Saint Louis, Missouri, United States) with activity 20,000 U per protein mg (activity 1 U is equal to hydrolysis of 1.0 microequivalent of fatty acid from a triglyceride in 1 hour at pH 7.7 at 37 °C using olive oil).Citric acid, sodium hydroxide, ethanol, and formaldehyde were obtained from Penta (Prague, Czech Republic).Te organic and inorganic reagents used in these experiments were of analytical grade.Deionized water was prepared using the Aqua Osmotic 02 device by Aqua Osmotic, Tisnov, Czech Republic.

Analyzed Samples
(i) Sodium citrate bufer 0.1 mol/L with a pH of 5 and 6, potassium phosphate bufer 0.1 mol/L with pH 7 and 8, and sodium phosphate bufer 0.2 mol/L with pH 9.0 were used as standard samples for the pH assay.Te saline served as a blank.Urine samples were used untreated.(ii) Glucose was solved in pH 7.4 and urine.Te solutions served as samples for the glucose assay.Phosphate-bufered saline (pH 7.4) served as a blank.(iii) Te solution of human serum albumin at pH 7.4 and the solution of human serum albumin in urine were used as standard samples for the protein assay.Phosphate-bufered saline (pH 7.4) served as a blank.(iv) Te porcine pancreas was solved in phosphatebufered saline (pH 7.4) and urine was used as an analyte, substituting leukocytes.Phosphate bufered saline (pH 7.4) served as a blank.(v) Urine from anonymized human volunteers was used for validation purposes.

Urine Test Strips Assay.
Standard urine test strips (DekaPhan Leuco, Erba Lachema, Brno, Czech Republic) were used for the experiment (photograph in Figure 1).A test strip contained 10 squares each to test one biochemical marker (specifc gravity, leucocytes, nitrite, pH, protein, glucose, ketones, urobilinogen, bilirubin, and blood/hemoglobin).Te strip was placed on a white paper surface, covered with the hollow part of the 3D-printed holder, and a smartphone (Redmi Note 11 Pro, Xiaomi Inc., Haidian District, Beijing, China) was located on the upper of the holder in a way as shown in Figure 1.After that, a sample of 10 μl was applied per square and incubated for 60 seconds, respectively, 120 seconds when leukocytase activity was measured and the test strip was photographed.Te smartphone camera was set to zoom 1×, automatic fash, and automatic white balance.Te camera was focused on the spot where the detecting square was placed, and the picture was collected in an 8 bit jpeg format.Five diferent test strips were analyzed for every sample.Te diference in color depth was calculated from the two photographs.

Measuring Color Depth.
Te value of color depth was measured by the software GIMP 2.10.34 (free and opensource software).Te spot with the square where the analysis took place was analyzed in fve randomly selected spots with a distance higher than 2 mm from the square´s edge.Te color depths were measured for the red (R), green (G), and blue (B) channels.Te fnal mean average color depth was determined.Because the 8 bit format of photographs in jpg format was acquired, the color channels can get a value between 0 and 255. 2 International Journal of Analytical Chemistry plastic disposable cuvettes and an Evolution 201 spectrophotometer (Termo Fisher Scientifc; Waltham, Massachusetts, United States) [15].(iii) Proteins were analyzed using the Bradford spectroscopic assay [16].Te Bradford reagent kit (Sigma-Aldrich) was chosen for the purpose, and the assay was performed in compliance with the protocol provided by the manufacturer.Te assay was performed using a standard spectroscopic assay using standard 1 cm plastic disposable cuvettes and a spectrophotometer Evolution 201.(iv) Te porcine pancreas was analyzed as a substance that mimics leukocytes and their leukocytes esterase.Indoxyl acetate served as a chromogenic substrate for an assay in standard 1 cm disposable cuvettes and the Evolution 201 spectrophotometer.
Esterase activity can be easily measured by spectrophotometry using indoxyl esters [17].In this article, the use of indoxyl acetate in the way previously described for cholinesterase and lipase assays was chosen [18][19][20].

Data Processing.
In the camera-based colorimetric assay, the average color depth (fve points randomly selected at a distance from the square side equal to 1/3 of the square side length) was calculated for a sample and a blank (matrix for sample solving).Te diference in color depths was calculated from sample and blank assays: Δ Color depth � Color depth (blank) − Color depth (sample).All samples were measured on a fve-time repeat.Te mean and standard deviation were calculated from repeated measurements.Te limit of detection was determined from the calibration curves using the rule that it is equal to the point in the calibration that numerically corresponds to the triplicate of the blank assay signal (rule S/N � 3).

Results and Discussion
In this study, assays of four markers typical for urinalysis by a strip test were chosen.Te pH assay represented an assay of an organic marker assay, the human serum albumin represented a protein marker, and the assay of lipase as a substitute for leukocytase esterase activity represented an enzymatic marker common in clinical biochemistry.
Te pH resulted in the construction of the calibration.Examples of test strip colorations and the calibration curves are shown in Figure 2. Te test strips changed colors from orange in acidic bufer pH 5 per light and dark yellow (pH 6 and pH 7) to green (pH 8) and fnally blue (pH 9).Te sensitivity of the pH assay was very low in the G channel while the R and B channels proved to be applicable.Both R and B channels were well correlated with the change in pH, as the R channel exerted a coefcient of determination of 0.998 and the B channel had a very similar coefcient of determination of 0.997.Te diference in color depth had a higher dynamic range for the B channel than for the R channel which also had a lower sensitivity for the acidic bufers.Te B channel appears to be optimal for practical use, and the pH based on R and B channels is sensitive enough to cover the physiological range of urine pH and serve to recognize pathologies connected with aciduria or alkaluria.
Te glucose assay was performed for the calibration range of 1.4-55 mmol/L and phosphate-bufered saline served as a blank.Examples of cuts from test strips coloring in the presence of glucose and the calibration curve are depicted in Figure 3. Te best sensitivity for these cuts from urine test strips was exerted by the assay in the R channel where the limit of detection of 0.11 mmol/L was reached for the glucose and the assay also had the highest dynamic range.Te assay in the G channel had a lower dynamic range and limit of detection for glucose was 0.60 mmol/L.Te worst result was observed in the calibration of glucose using the B channel where a limit of detection equal to 1.8 mmol/L for glucose was achieved and the dynamic range of color depths was the lowest one.Te inverse proportionality of the calibration in the B channel is another notable fact.Nevertheless, calibrations in all three color channels were sensitive enough to provide limits of detection under the value of the renal threshold for glucose, which is around 10 mmol/ L depending on the exact type of pathology [21,22].
Te human serum albumin solution was analyzed for a calibration range of 0.3 -5 g/L and phosphate-bufered saline served as a blank.Te colors of the cuts from test strips as a result of the protein assays and calibration for human serum albumin are depicted in Figure 4. Limits of detection of 0.15 g/L for the R channel, 0.24 g/L for the G channel, and 0.22 g/L for the B channel were calculated from the calibration.Te coloration of the zones for the proteinuria assay by the urine test strips came from yellow (no proteins) to green (maximal concentration of the analyzed protein).Te color change was similar to the zones for the glycosuria assay, but while the glucose assay provided contrast color change, the assay of proteins led to a light green, and the dynamic range of the change in color depth was lower.While the change in color depth reached up to 200 for the glucose assay and R channel, the assay of proteinuria reached around 70 in the best R channel.Te other channels exerted even lower sensitivity.
Porcine lipase which served as a surrogate for leukocytase esterase is not a common commercially well accessible and the test for leukocyturia was verifed this way.Examples of the assay and calibration of the urine test strips for lipase are presented in Figure 5. Te part of the urine test strip for leukocyte detection provided quite a weak color change from colorless to a bright violet.Te weak coloration resulted in a low dynamic range of the change in color depth.Te best response to the color change was observed in the R channel, which was equal to approximately 50; the G channel had the maximal color change around 45; and the B channel around 10. Te limit of detection for the lipase assay was equal to 2.50 U/μL for the R channel, 3.40 U/μL for the G channel, and 24.2.for the B channel.Te activity of lipase cannot be easily recalculated to the number of lymphocytes, but an activity of 50 U/μL respectively, 500 U per a standard sample 10 μL is roughly equal to 500 lymphocytes considering the color etalon provided by the test strip manufacturer.International Journal of Analytical Chemistry Te aforementioned assays were validated using intact urine samples (pH measurement) or spiked (glucose added in the glucose assay, human serum albumin added in the protein assay, and porcine lipase added in leukocyte assays).Te B channel for the pH assay and the R channels for the remaining three assays were used for the analysis of photographs from the assays by urine test strips.Te exact value of the measured parameter was determined from the calibrations above using urine test strips or by standard assays calibrated using the same standards for the calibration.Te results of the validations are summarized in Table 1.When comparing the data reached by the use of urine test strips and the potentiometry for pH and spectroscopy for glucose, human serum albumin, and lipase assay.In the comparison, there were no signifcant differences by ANOVA at the probability level of 0.05.Te selected markers and their expected concentrations that can occur in the urine can be measured with urine test strips with a smartphone camera as a detector quite easily and provide data for diagnosis with similar accuracy as the standard methods.

International Journal of Analytical Chemistry
Te use of a combination of a smartphone camera and urine test strips, as presented here, falls within the scope of the idea of point-of-care urinalysis with a simple detector device.Instrumentation for point-of-care urinalysis tests has been extensively reviewed [23][24][25][26].Several biosensor applications for particular markers in urine were also presented in recent journal articles [27][28][29][30][31].   International Journal of Analytical Chemistry

Conclusions
Urinalysis by standard urine tests has limitations in the low reproducibility and limited quantifcation of the analyzed markers.Te assay can, however, be improved by a colorimetric sensor that can even be a smartphone camera.Pointof-care tests based on this combination remain inexpensive and easy to perform but the subjective scaling of coloration by the human eye is replaced by the smartphone camera.Moreover, such an assay is quantitative and can provide results and support for precision diagnosis similar to standard laboratory methods.Te urine test strips can be read by the naked eye as intended by the manufacturers; however, this method provides results highly dependent on individual abilities, sensitivities to distinguish color, and experience with such tests.Te combination of smartphone and urine test strips reduces the mentioned drawbacks without raising additional costs, requiring sample processing, or introducing an elaborate assay procedure.Te fact that a cheap smartphone camera integrated even into a common device is sufcient for the analysis makes the whole assay more competitive.Te necessity to buy even a cheap analytical device can be a substantially limiting factor for the distribution of a point-of-care test.Because the assay needs only a common smartphone without any functions above standards, access to the improved method here is unlimited.

Figure 2 :Figure 3 Figure 3 :Figure 4 :
Figure 2: (a) Examples of urine test strips that are sensitive to pH and exposed to various bufers with pH indicated under each picture.(b) Use of urine test strips for bufers with various pH.Each sample was measured fve times, and error bars indicate the standard deviation.Tree curves were made for each color RGB channel.

Figure 5 :
Figure 5: (a) Examples of urine test strips cut for leukocyte assay exposed to various activities of porcine lipase are indicated under each picture.(b) Use of urine test strips for leukocyte assay, represented by measurement of porcine lipase as a surrogate.Each sample was measured fve times, and error bars indicate the standard deviation.Tree curves were made for each RGB color channel.