Use of RGB Color Sensor in Colorimeter for better Clinical measurement of Blood Glucose

Transcription

1 Use of RGB Color Sensor in Colorimeter for better Clinical measurement of Blood Glucose A.Sivanantha Raja, 2 K.Sankaranarayanan Alagappa Chettiar College of Engineering and Technology, Karaikudi , India. sivanantharaja@yahoo.com 2 V.L.B.Janakiammal College of Engineering and Technology, Coimbatore , India. kkd_sankar@yahoo.com Abstract The Gold standard for testing blood glucose in Colorimeters is the measurement of glucose in a plasma sample obtained from a vein. This method involves a chemical reaction activated by an enzyme called Glucose Oxidase. Most of the Colorimeters used for the Glucose measurement in Clinical laboratories perform well in the mg/dl range of blood glucose. In Self Monitoring Blood Glucometers (SMBG), a drop of blood is placed on a small window in a test strip. Blood glucose acts as a reagent in a chemical reaction that produces a color change. The color change is detected by a reflectance-meter and reported as a glucose value. They too perform well in the range of mg/dl range of Blood Glucose. Moreover, requirement of special purpose Dry slides is a big problem for the users of Glucometers. This paper aims to sandwich the technologies of Colorimeter and SMBG and interface the Clinical instrument with the Computer for a better clinical diagnostic result. The existing Colorimeter is upgraded with RGB Color sensor and Microcontroller and interfaced with the Computer to make Clinical measurement easier. Experimental results show that this modified Colorimeter can perform better than the existing colorimeter. Keywords: Clinical Instruments, Blood Glucose measurement, Colorimeter, Color Sensor, Color measurement, hue and saturation.. Introduction Blood Chemical tests are done at the Clinical Laboratories to help Doctors for making a prognosis. Hence error in the measurement of concentration of Blood Chemical would misguide the Physicians and that would result severely in the treatment of patients. In the case of diabetic patients, estimating their blood sugar is important for avoiding hypoglycemia and hyperglycemia - blood sugar lows and highs. Controlling the Blood Sugar of Diabetic patients is their number one goal. For regular monitoring, Self Monitoring Blood Glucose (SMBG) Meters are used. In Self Monitoring Blood Glucose (SMBG), a drop of blood is placed on a small window in a test strip. Blood glucose acts as a reagent in a chemical reaction that produces a color change. The color density is measured by a reflection spectrometer. A mathematical function is stored in the analyzer s central processor is used to convert the measured dye density into analyte concentration. This mathematical function is determined during calibration. Gold Standard Blood Glucose test is done in the Colorimeter as per Glucose Oxidase method [,2] in the Clinical laboratories. The complex substances appearing in Blood serum develop Color following the reaction with a reagent. Colorimeter measures the concentration of the Blood substance based on the amount of absorption of a monochromatic light in that solution. Concentration of Blood Glucose in an unknown assay is estimated by comparing the Absorbance value measured in the Colorimeter with that of the Absorbance measured for known concentrated Glucose solution. SMBG Meters perform well only in the range of mg/dl range of Blood Glucose. [3,4,5]. They need to be calibrated once in a month or whenever a new set of Dry slides used for Glucose measurement [6,7,8]. However, Gold Standard test for Blood Glucose measurement must be carried out in Colorimeters only. An error of 5 20% is not uncommon in colorimeters because those measurements require a highly monochromatic source [9,0]. The spectral bandwidth of the filters used for this purpose and the operational wavelength of the photo detector are also behind the inaccuracy in the result. Colorimeter exhibits more measurement complexity especially in the non-linear region of the reagent solution. For example, only up to the Glucose concentration of 400 mg/dl, the reagent shows linearity in color change. Error in the measurement of Blood Glucose in the high Glucose level i.e., Hyperglycemia and low Glucose level i.e., Hypoglycemia might result in the perennial complications such as Heart attack, Brain attack, Strokes, Blindness, and Kidney failure, etc.[]. Hence error should be minimized in the Clinical Laboratory equipments. 23

2 Objective of this paper is to modernize the Colorimeter by using RGB Color sensor that is interfaced to the Digital Computer [2]. Estimation of the analyte concentration is to be done as that of SMBG meters using RGB data received from the color sensor instead of Absorbance. The RGB data is to be converted to its equivalent hue and saturation as per the Chromaticity Diagram [3 6]. From the saturation of the Color, the concentration of the Blood Glucose can be estimated. This proposed method is superior in two aspects. One is the non-requirement of monochromatic source in the Colorimeter and the second one is the automation of Clinical instrument. 2. Theory The theories of the existing and proposed Colorimeters are discussed below. 2. Existing Colorimeter The instruments used for the Blood Glucose concentration measurement are not Colorimeters but Absorption meters, since it is the meter that measures the amount of light absorbed in the Color concentration of the solution (Figure.). After setting the initial condition in the Colorimeter, 0µL of the standard Glucose solution with the concentration of 00 mg/dl (C Standard ) is added to ml of the blank reagent solution in the test tube and it is kept inside the incubator at the temperature of 37 c. After 0 minutes, Purple Color would have developed in the assay. This test tube is inserted into the Colorimeter and the monochromatic source of wavelength 520nm is passed through the assay. The Absorbance shown on the display of the Colorimeter is noted as A Standard. Then, 0µl of the unknown concentrated Blood Glucose with concentration C Unknown is added to another ml of Blank reagent and kept at 37 c for 0 minutes. This assay s Absorbance from the Colorimeter is noted as A Unknown. This is substituted in equation () to find out the unknown Blood Glucose concentration C Unknown. A Unknown C Unknown = C Standard x () A Standard 2.2 Self Monitoring Blood Glucometers Multi-layer thin film dry slide technology is adopted in most of the handheld Glucometers. A Colorimeter slide (Figure.2) comprises several layers coated on a transparent plastic support. Fig. : Existing Colorimeter The Color concentration of Blood Glucose in the assay is accomplished electronically by detecting the Color light intensity passing through a sample containing the reaction products of the serum and reagent as per Beer-Lambert Law. A Yellow urine sample, for example, passes yellow light and absorbs blue and green. For this reason, and to obtain purity in measurement, optical Color filters are used to select a narrow wavelength spread (Bandwidth) of light that shines on the Photo detectors. Standard Glucose solutions are used for conducting experiments and the reagent used for the measurement of Glucose concentration is from AUTOSPAN. The basic assay parameters of this reagent are listed in Table. Table : Assay parameters Wavelength (nm) 505 Sample Volume (µl) 0 Reagent Volume (µl) 000 Incubation time (min) 0 at 37ºC Linearity High (mg/dl) 500 Stability of Color Hour Standard Glucose 00 mg/dl Absorbance Limit (max) Optical Path length (cm) Fig.2: Colorimeter Dry Slide Courtesy: Johnson & Johnson The sample is applied to the top or spreading layer. This ensures its uniform dispersion. It is also a micro filter, removing small particulate material and large molecules, such as proteins. It also contains a white diffuse reactant which acts as a reflecting medium during colorimetry made through the transparent support. Various underlying layers have different functions. They include a reaction layer which usually contains an enzyme catalyst and a registration layer where colored dye is formed in proportion to the analyte concentration. The color density is measured by a reflection spectrometer. Other layers may include physical barriers such as semi permeable membranes and scavenger layers which chemically alter potentially interfering endogenous substances. Interferences are thus minimized. For each test a mathematical function is stored in the analyzer s central processor. It is used to convert the measured dye density into analyte concentration. The function is a third degree polynomial. Its three coefficients are determined during calibration. 24

3 2.3 Proposed Colorimeter The drawbacks in the existing Colorimeters are the requirement of a highly monochromatic source and its inability to operate in the non-linear region of the assay. Both can be eliminated if the Color of the assay is directly used for Glucose measurement using Color measurement technique instead of measuring Absorbance. Interfacing this modified Colorimeter with the Computer is done for automatic processing as in the SMBG. In order to accommodate this technique, the existing colorimeter is modified as shown in Figure Color measurement technique From the 3 variable RGB data, Color s hue and saturation can be obtained using Chromaticity diagram. Since Color of the assay is constant irrespective of the concentration of Glucose, saturation alone can be used to relate the Color concentration and Glucose concentration. In place of saturation, the word purity is frequently used. 3. CIE 93 Chromaticity Diagram A diagram which is convenient for Color measurement is the Horse Shoe shaped Chromaticity diagram [3, 4]. The positions of the various Colors, from Blue (400nm) at one end to Red (700nm) at the other end, are indicated on the curve. Any point not actually on the solid line curve, but within the area enclosed by the curve, represents not a pure spectrum of Color, but a mixture of spectrum Colors. As White is such a mixture, it too lies within this area, specifically at point C (Figure 5). The most intense Colors i.e., Red, Green and Blue, in their deepest shades, are obtained at the outer edge of the diagram. The more familiar Color shades such as Pink, light green, pale blue appear when moved towards the centre. Finally at the centre comes the White point C. Fig. 3: Proposed Colorimeter The filter is removed and hence the monochromatic source is eliminated. In the place of Photodiode an RGB Color sensor Agilent HDJD-S722-QR999 color sensor as shown in Figure 4 is used to detect the presence of a certain color and identify its exact coordinate across the full color spectrum. Agilent's color sensor converts colored light to proportional RGB voltage outputs. The voltage output of each channel (R, G and B) increases linearly with increasing light intensity. Fig. 5: CIE 93 Chromaticity Diagram Fig. 4: RGB Color sensor - Agilent HDJD-S722-QR999 This RGB voltage output is connected to the PIC 8725 Microcontroller that converts these analog voltages to three 8-bit Digital data. This 24-bit RGB data is linked to the Computer through the 9-pin Serial port. The data thus obtained in the Computer is processed as per the Color measurement technique to display the Glucose concentration. The term hue is associated with the dominant wavelength. Thus, hue represents the basic Color as it appears to us while saturation tells how deep the Color is. It is possible to specify the saturation of a Color on basis of its distance from Illumination point C. Chromaticity values depend on only dominant wavelength and saturation, and are independent of the amount of luminous energy. For example, a brown color is not on the diagram; however, a brown is just a low-luminance orange-red. A standard white light (formally called illuminant C) is located near (but not at) x=y=z=/3. If two colors are represented by points C and C2, the additive mixture is a point C3, lying somewhere on the line CC2. Complementary colors are those that can be mixed to produce white light. If a line is drawn from the white point through a point representing a specific color, the intersection of this line with the spectral locus defines the dominant wavelength. If the line crosses the purple 25

4 boundary, the dominant wavelength cannot be defined, and such color is called non-spectral. Indeed, there is no magenta in the rainbow. 3.2 Conversion of RGB data to hue and saturation Conversion of RGB to XYZ takes the form of simple matrix transformation [6]. Chromaticity coordinates x, y and z are obtained from the Tristimulus values X,Y and Z. X Y = Z R G B (2) X x = ; Y ; Z y = z X + Y + Z = (3) X + Y + Z X + Y + Z Hence, conversion of RGB data to its equivalent Chromaticity coordinates can be done using the above two equations. The Chromaticity coordinates that were adopted by the International Commission on Illumination for the various spectrum Colors are given in Table 2. Table 2 : CIE 93 Chromaticity coordinates Chromaticity coordinates Wave length [nm] x Y z Slope of the Color Slope of the line gives the value of Hue and this Slope of the Colors with respect to the luminance point C in CIE 93 Color space is calculated using equation (4) if the Chromaticity coordinates x, y and z are known. Slope = (y white y) / (x white x) (4) where x white = y white = /3. Hence, hue can be measured. It is possible to specify the saturation of a Color on the basis of its distance from point C. 3.3 Mathematical model For measuring the concentration of Blood Glucose using the RGB data, the Color Saturation (x) values obtained for various known concentrations of Glucose (y) are fit into an m th order polynomial equation and this equation will characterize the behavior of reagent used for Blood Glucose measurement. In general for any order say m, the polynomial equation will be, m m 2 y = a (5) m x + am x a2 x + ax + a0 From this equation, (m+) simultaneous equations will be generated. Solving these (m+) simultaneous equations (using Cramer s rule), the value of the unknowns a m to a 0 can be calculated. Equation () used with Colorimeter is suitable only for linear responses whereas this polynomial equation of the form of equation (5) is suitable even for non-linear responses. This polynomial equation should be the characteristic equation of the reagent used for Blood Glucose measurement. Once this polynomial equation is ready then the Saturation (x) of the Unknown Glucose concentrated assay can be substituted in this equation (5) and the Unknown Glucose concentration (y) can be estimated. 4. Experiment and Result Modified Colorimeter has been designed and its performance has been tested in the Laboratory as against the existing Colorimeter. Glucose standard solutions are used to conduct experiments in the laboratory. Glucose standard solutions of concentrations from 50 mg/dl to 800 mg/dl at the Glucose concentration gap of 50 mg/dl are taken and 2 tests in each category are conducted. Averages of the readings observed in those experiments in both old Colorimeter and new Colorimeter with Color sensor are tabulated in Table 3. This shows the response characteristic of the reagent used for the measurement of concentration of Blood Glucose. Glucose Concent- Ration [y] (mg/dl) Table 3: Characteristic response of reagent Existing Modified Colorimeter [x] Colorimeter Saturation Absor- (%) bance R G B (O.D.) Hue (nm) c c c c c c c c c c c c c c The graph in Figure 6 shows the responses of Glucose concentration Vs Absorbance measured in the existing Colorimeter and Glucose concentration Vs Purity of the Color measured from the RGB Data of the proposed Colorimeter. Hue of the assay is reasonably constant for all glucose concentrations. 26

5 Absorbance (O.D) Response curve of Reagent Saturation (%) software does the RGB data to its equivalent Purity or saturation of the color and the dominant wavelength of the color as explained in chapter 3. using the CIE 93 Chromaticity diagram. By choosing the Check for a color button on the screen, the RGB data displayed in the LCD screen of the modified Colorimeter can be entered. Figure 8 shows the screen of such RGB data entry. Glucose Concentration (mg/dl) Absorbance (O.D) Saturation (%) Fig. 6: Response curve of Reagent used for Glucose measurement This graph clearly shows the nonlinearity in the response of reagent that is used for the measurement of Glucose. Non-linearity prevails in both Absorbance and Saturation when the Glucose concentration is above 400mg/dl. As in the SMBG, the Glucose Concentration (y) versus the Saturation (x) data in the Table-2 is fit into a polynomial equation of degree 6. This equation is used for the actual measurement of Blood Glucose using the proposed Color sensor based Colorimeter. The equation for Table-2 is y = x x x x x x (6) with Correlation Coefficient R = This equation (6) characterizes the behavior of reagent used for Glucose measurement and the equation s correlation coefficient is which is very accurate. This procedure can be repeated whenever the reagent is changed or whenever the calibration is required for the instrument. Once this equation is ready, the unknown concentration of the Blood Glucose can be estimated from the RGB data and its saturation value x substituted in the equation (6). This can be done easily in the Computer. In order to estimate the Percentage Error in the measurements of Glucose concentration in both Existing Colorimeter and the proposed Colorimeter, experiment is again conducted with Glucose standard solutions by using the equations () and (6) respectively. The responses are shown in Figure (7). Error (%) % Error in Glucose measurements Glucose concentration (mg/dl) Fig.7: Comparison of Error in Glucose measurements Figure (7) shows the steep increase in the measurement Error while using the Existing Colorimeter, that too severely, in the non-linear region of the assay. The proposed Colorimeter offers less than 0% of error even in the non-linear region of the reagent. 5. About Software The software for the Color measurement has been developed in Visual Basic by the authors. This Fig.8 RGB data entry When Ok button is pressed the software will display the RGB data s equivalent Wavelength of the Color and its purity as shown in Figure 9. Fig.9 RGB data s equivalent wavelength and purity The data thus collected are entered in MS- EXCEL sheet and the graph is plotted between the concentrations of Glucose (mg/dl) versus purity of the color. When the graph is plotted, using the Add trend line option available in the Excel, the polynomial equation of 6 th order (Equation 6) along with its Correlation coefficient (R) is found. This polynomial equation is used for further measurements. 6. Conclusion As the Photo detectors used in Colorimeters offer their efficient performance in the wavelength of 520nm (Green), usually its complementary Color Pink or Purple is chosen as the Color developing in the assay. But, this proposed system does not have such Color restrictions. Moreover this proposed method overcomes the shortcomings in the existing system such as requirement of Monochromatic source and the Nonlinearity in the density of Color developed in the assay. 27

6 The hurdle in the use of special purpose Dry slides in the SMBGs is also overcome in this proposed method. A dry slide meant for one Glucometer cannot be used with other Glucometer because the Dry slide s performance curve equation would have been stored in the Processor of the Glucometer. This equation can never be changed and also cannot be used for other slides. Whereas, in the proposed Colorimeter, calibration can be done in the Clinical Laboratory itself by the Lab technicians whenever required and this function can be stored in the Computer for further use. Hence any reagent can be used after calibration. Software for this purpose is also developed by the authors. 7. Future trend As this work is new of its kind, further work can be done to improve the software so that many of the processes can be done automatically. The serial interfacing provided in the Modified colorimeter has not been fully utilized in the present software. New software may be developed to work with the RGB data coming straight from the modified colorimeter to the computer. The entire work has been done with the Glucose standard solutions only. Further work can be carried out in the Clinical laboratories with real blood samples collected from the Diabetic patients and the result of this Colorimeter can be compared with that of the result obtained from the existing colorimeter and/or Glucometer. 8. References [] David T Plummer: An introduction to Practical Biochemistry, Tata McGraw-Hill, 3/e, 998, pp [2] Harold Varley: Practical Clinical Bio-chemistry, CBS Publishers and Distributors, 4/e, 998, pp. 9-24, [3] Paul L.Edmiston, Theodore R.Williams: An analytical laboratory experiment in error analysis: Repeated determination of Glucose using commercial Glucometers, Journal of Chemical Education, Vol.77, No.3, Mar [4] Vallera Da, Bissel M G, Barron W: Accuracy of portable Glucose monitoring. Effect of Glucose level and prandial state, Journal of Clinical Pathology, Feb 95(2); [5] Robbert J Slingerland. Kor Miedema: Evaluation of portable Glucose meters. Problems and recommendation, Journal of Clinical Pathology, Sep 2003; [6] Carol Rados: Safety Alerts on Blood Glucose meters, FDA Consumer, Proquest Science Journals, Vol. No. 40:2, March/ April 2006, pg.6. [7] James A Fain: Blood Glucose Meters: Different strokes for different folks, Nursing, Proquest Science Journals, Vol. No.34:, Nov 2004, pg.48. [8] Monika Mehta, et al.: Emerging technologies in diabetes care, U.S.Pharmacist, Vol No.27:, Nov [9] Joseph J.Carr and John M.Brown: Introduction to Bio-medical Equipment technology, Pearson Education, 4/e, 2003, pp [0] R.S.Khandpur: Handbook of Bio-medical Instrumentation, Tata McGraw-Hill, 2003, pp [] A R Feinstein: Misguided efforts and future challenges for research on diagnosis tests, Journal of Epidermy & Community Health, Dec 200; 56: [2] Sivanantha Raja,A and Sankaranarayanan.K : Measurement of Concentration of Bio-chemical components using Color image processing, Proceedings of the National Conference on Biomedical Engineering (NCBME) at GITAM,Vizag, India in December [3] D.L.Macadam: Color Measurement, Theme and variations, Springer series in Optical Sciences, 2/e, 985, pp [4] R.R.Kulati: Monochrome and Color Television, Wiley Eastern Ltd., 992, pp [5] Y.E.Khoroshaylo, et al.: Colorimetry, CAOL, Yatta, Crimea, Ukraine, Sep 2005, p276. [6] Laser Components Instruments Group Inc., Product information on Color measurement with Color Sensors, A.Sivanantha Raja, born in 966, graduated in Electronics and Communication Engieering from the Madurai Kamaraj University in 987. He did his Master programme in Microwave and Optical Engineering in 997 in the Madurai Kamaraj University. He has been working as Senior Lecturer in Alagappa Chettiar College of Engineering and Technology, Karaikudi, Tamilnadu, India since 988. His areas of interest are Microwave Integrated Circuits; GPS based applications and Biomedical Engineering. Currently he is undergoing his research programme in Anna University, Chennai (Madras) under the guidance of Dr.K.Sankaranarayanan, the co-author of this paper. K.Sankaranarayanan, born on , completed his B.E. (Electronics and Communication Engineering) in 975 and M.E. (Applied Electronics) in 978 from P.S.G. College of Tech., Coimbatore under University of Madras. He did his Ph.D. (Biomedical Digital Signal Processing and medical Expert System) in 996 from P.S.G.College of Technology, Coimbatore under Bharathiar University. His areas of interest include Digital Signal Processing, Computer Networking, Network Security, Biomedical Electronics, Neural Networks and their applications, and Opto Electronics. At present he is working as Dean of Electrical Sciences at V.L.B.Janakiammal College of Engineering and Technology, Coimbatore, Tamilnadu, India. 28