1 Abstract—This Paper studies the methodology that utilizes optical transmission spectroscopy techniques for the non-invasive measurement of blood glucose concentrations in diabetic patients. Mid-Infrared (Mid-IR) range wavelength was used to study the variations of glucose concentrations in two different media: saline-blood and pure blood samples. Results obtained from both categories of samples were analyzed and presented in this paper. I. INTRODUCTION Diabetes is a condition faced by an individual when his/her blood sugar levels are abnormally high or low. There are mainly two types of diabetic patients, although having the same effect they imply due to different reasons. Type 1 diabetes has its adverse effects due to lower insulin production in the body. However, type 2 diabetes is due to the fact that the body organs fail to respond to insulin, a condition also known as insulin resistance. This condition doesn’t allow the sugar to enter into blood thereby increasing the level of sugar in the blood as well as the urine. This is also known as hyperglycemia [1]. Type 1 and type 2 diabetes were identified in 1935, and that was considered a breakthrough in the study of diabetes at that year [2]. This leads to a whole new study in this particular field. In 1960’s urine strips were developed, and in 1970’s insulin pumps were developed to assist the patients with continuous injection of insulin. Many other sophisticated instruments were devised later in the 80’s and 90’s, some proved to be useful where as others didn’t sometimes due to accuracy, cost or failing to be user friendly. While glucose has a close a relation to all parts of the body, it has a unique relationship with the nervous system, specifically, with the brain. We are trying to find out if a relation between the functions of the brain and the level of glucose in the bloodstream can be derived so that glucose can be measured using the brain in a non- invasive manner. The unique fact about the brain is that unlike other organs, it does not require and consume glucose in a regular or timed manner, so the brain activity and glucose intake does not guarantee causation. However, the hypothalamus part might help us find a causation relation between its activity and glucose level in the blood since the hypothalamus is more closely related to the endocrine system than other parts of the brain. So, it contains the neurons that detect the amount of glucose present in the blood [3]. However, analyzing the hypothalamus is made more difficult since it’s smaller than a marble and located close to the center of the skill. Implications of this will be made clearer later. Several techniques have been used to determine the glucose level noninvasively, but most methods are expensive, difficult to be used, high cost and lack of selectivity. Below are the most common noninvasive techniques used: 1)The Electroencephalography (EEG): EEG has a very high temporal resolution (around 1 millisecond) but a relatively low spatial resolution when compared with other tests that measure brain activity, such as MRI or CT [4]. Detecting hypothalamus signals using EEG is extremely difficult for the following reasons. Firstly, maintenance of blood glucose levels is an autonomic function. In other words, it does not require the person to think about hormones to trigger hunger as well as conversion between glucose and lipids. Second, its size and position makes it very difficult for the EEG to pick up any generated signals. 2) MRI and fMRI: For detecting the activity of the hypothalamus, the fMRI may be more suited because unlike EEG, it can record signals from all of the parts of the brain as opposed to those parts closes to the skull. While fMRI techniques regarding the function and activity of the hypothalamus in relation to blood glucose level appear promising there are still some obstacles to be tackled [5]. Firstly, the hypothalamus itself may not be enough to be analyzed to find such a correlation; a third variable, such as hunger or kidney activity may need to be analyzed at the same time to make the relationships clearer [6]. Secondly, the cost of MRI machines crosses at least a half million USDs, so while tests can be done to find a correlation or relationship, using it as a primary tool for measuring glucose at this point remains largely unpractical. 3) Raman Spectroscopy Raman spectroscopy is a relatively new method for measuring blood glucose non-invasively using mid- infrared or near-infrared light [7]. The technique uses the Non-invasive Blood Glucose Measurement Using Transmission Spectroscopy Yahya Khawam, Muddassar Ali, Haris Shazada, Sofian Kanan, Hasan Nashash American University of Sharjah,
[email protected] 978-1-4673-2821-0/13/$31.00 ©2013 IEEE 2 principle that light reflected and refracted from the blood will do so in a manner different based on the level of glucose present, allowing measurements to be taken. Since it is a fairly new technique, there are still many obstacles to be tackled. One of the obstacles is that near-infrared light penetrates only half a millimeter below the skin [8], so that glucose level is measured in the fluids in the skin rather than directly in the bloodstream. Researchers are trying to get around the problem by finding and relating the different concentrations to accurately predict the level in the blood. Nutrition intake is another problem since while the glucose level in blood changes rapidly the level in the layers of skin can take as many as 10 minutes to show the same change. To get more accurate results, the researchers are including the rate of diffusion of the glucose from blood to skin in their calculations [9]. While this research appears promising, the accuracy levels are not on par with invasive techniques. II. SIGNAL PROCESSING TECHNIQUES USED The specific method of light spectroscopy technique that was used is transmittance. When using light spectroscopy in the transmittance mode, the two main methods for quantitative analysis are based on measuring the peak height and the peak area. The peak height method simply takes the height of a peak that corresponds to the substance as a measure of the amount of a particular substance present in the entire sample. The integral method however takes the area under a defined peak to determine the concentration of a given substance in the sample which is more reliable method for quantitative measurements in spectroscopy. Since the peak intensity takes the area of the peak relative to the entire spectrum, it is not affected by the thickness of the blood sample. III. METHODOLOGY AND RESULTS This research project made use of real blood samples in order to look for the relations between the light transmitted and the glucose concentration in blood samples. However, blood samples were not obtained randomly meaning that there was certain division that must be kept in all samples that is between 10 mg/dl up to 20 mg/dl. Spectra were recorded for saline-blood and pure blood samples in the full spectral range. A. Spectrum of Saline-Blood Samples in Micrometer Range In medicine, saline (also saline solution) is a general phrase referring to a sterile solution of sodium chloride in water but is only sterile when it is to be placed parenterally (such as intravenously). Otherwise, a saline solution is simply a salt water solution [10]. The use of saline is to avoid the blood from being clotted and thus control the glucose concentration in the blood. To illustrate more, variant blood samples with various glucose concentrations were made using only one blood sample, called base sample that has an initial glucose concentration, typically the lowest. Then dilute saline with saline-glucose mixture that has fixed glucose concentration. This will help control the glucose concentration in the new diluted saline-glucose samples. The next step is to mix the diluted saline-glucose concentration with the base sample and then calculate the new concentration in the saline-blood samples. In this way, variant blood samples can be obtained with each sample has different glucose concentration. Figure 1 shows the spectra of various saline-blood samples. As shown in Figure 1, the presence of water in the solutions made it difficult to monitor the full range of spectrum. Therefore, only two regions can be used to quantify the amount of glucose in the given sample namely, from 3200-1670 and 600-1300 cm-1. In order to see the relation between light transmission and glucose concentration in blood, a calibration plot should be conducted. However, this relation cannot be directly seen from saline-blood spectrum. The reason behind that is because of the sampling process where a drop of saline-blood sample was placed on the top of ZnSe window. Then saline-blood drop will be spread around the window. As a result, the amount of spreading cannot be the same all the time meaning that it is difficult to control. The thickness of the saline-blood sample seen by the beam of light is always different and thus causes scattering of the incident light. As a result, there will a different shift for each sample's spectrum. Figure 1. FTIR spectra of various blood saline samples. 3 Figure 2 shows the effect of change in the drop's thickness in the window. In this figure, the spectrum of two saline-blood with the same glucose concentration, but with different thickness labeled as high and low. Figure 2. Effect of the change in volume on the ZnSe window As can be seen from, the saline-blood sample with the low volume has more transmission since it is more transparent than the sample with large volume. In order to eliminate this effect, a technique called the "integral method" was used. The integral method calculates the area under the peaks where the limits are chosen between two points at which the features of the peak for each saline-blood sample is similar but varied based on the glucose levels. After obtaining the values of the areas under each peak for each sample, a calibration plot was made to see the relation between the light transmission in blood and glucose concentration. The calibration plot indicates that the sampling of the blood saline solutions is not consistent. B. Spectrum of Pure Blood in Micrometer Range Unlike the methods described before, glucose concentration in this experiment was in natural increments with no chemical compounds added to the blood samples. The steps of obtaining the blood samples are different than the first experiment. To illustrate further, participant who wish to provide blood, he/she must take blood sample while fasting, and then eat food that contains good amount of glucose. Next the person needs to keep a track of the glucose concentration using the invasive glucose-meter, and whenever there is appropriate change in glucose concentration, 10-20 mg/dl, a blood sample is to be extracted. Moreover, the glucose concentration in the participant’s blood was tracked with respect to time. Figure 3 shows the spectrum of five different pure blood samples. Figure 3. FTIR spectra for various blood samples with different glucose concentrations A calibration curve was made using the integral method (see Figure 4) Figure 4. Peak Intensity vs. glucose concentration As can be seen from the calibration plot, two frequency bands were used to represent the relation between light transmission and glucose concentration. In the wavenumber range 1196-1146 cm-1, the relation is: T = 31750݁ି.ଷଶ, whereas the relation between light transmission and glucose concentration in the range 1146-1009 cm-1is: T = 83171݁ିଷ Where T is the transmission of light, and C is the glucose concentration. In wavenumber ranges, the accuracy and the sensitivity were measured and summarized in Table 1. TABLE 1 MATHEMATICAL MODEL comparison 1196-1149 cm-1 1146-1009 cm-1 Rcorrelation (R2 ) 0.8205 0.8249 Accuracy 93.3% 93.2% Sensitivity 1006.4 2495.13 Correlation between both bands 0.99 4 C. Analysis of the Results: In the first experiment, spectra of saline-blood samples were obtained. However, using such samples, no conclusion can be made about the detection of glucose in the saline-blood samples. Therefore, the second experiment was performed using pure blood samples. After obtaining multiple spectra for variant- glucose concentrations and normalizing the spectrums using the integral method, a negative-exponential trend between peak intensity and glucose concentration was obtained in the wavenumber range 1196-1146 cm-1 and 1146-1009 cm-1. A comparison was made between both ranges to see which best range to use in order to calculate glucose concentration. As shown in table 1, the R2 value was calculated to be 0.8205 and 0.8249 for ranges of 1196-1146 cm-1 and 1146-1009 cm-1, respectively. These values show that the model, peak intensity vs. glucose concentration, is reasonably acceptable to predict other glucose concentrations. However, for both ranges, the accuracy is almost the same, 93.3% and 93.2% but their sensitivity values are varied. Even though both ranges have different sensitivities to glucose concentrations, they both correlated and it is 0.99 since both them have the same behavior. Despite all results obtained in this project, there is major limitation in using the method of spectroscopy to calculate the glucose concentration in blood samples. Generally, it is known that molecules with O-H bonds absorb the infrared radiation energy in the wavelength range studied. Since glucose and other molecules have such bonds, there is varying levels of contribution to the peak intensities for the various glucose concentrations from other molecules. After reviewing transmission database written by Silverstein for various molecules, there is transmission for other molecules in the range that was chosen for this project to measure glucose concentration. The transmission for many molecules that overlap with the project’s range, 1196-1146 and 1146-1009, seen from the database varies from 19-95% transmission [11]. As a result, when implementing the third phase of the project which is devising the model, the researcher should avoid the points where the transmission for other molecules is low to ensure lower error when measuring glucose concentration. IV. CONCLUSION It can be concluded that optical spectroscopy is a viable methodology to detect the glucose levels in the blood. It uses a range of wavelengths of near infrared light to detect different concentration levels of glucose in blood. Moreover the results also provide a strong correlation coefficients which prove the fact that there is a strong relation between the data obtain within different range of wavelength of infrared spectroscopy, yet there is still major limitation discussed that introduces an error in the system's model. The future work of this project is to device a system that has the ability to non-invasively measure glucose concentration in diabetic patients using the principle of optical spectroscopy. The system devised will be implemented such that it meets certain criteria such as: cost, time required for measurement, accuracy, overall size and ease of use. Lastly, this project serves a very important purpose of setting the parameters to devise in future a non-invasive system to measure glucose levels in the blood. REFERENCES [1] Steven Edelman. “What is the difference between type 1 and type 2 diabetes.” abcnews.go.com, Aug. 12, 2008. [Oct. 17, 2011]. [2] C.Best et al. “The preparation of insulin.” The Journal of Biological Chemistry.[online]. 57(1), pp. 709-723. Available: www.jbc.org [Oct. 12, 2011] [3] Denis Burdakovet.(2005, Nov.).“Glucose-sensing neurons of the hypothalamus.”Philosophical transitions of royal society. [online]. 360(1464), pp. 2227–2235. Available: [Feb. 26, 2012] [4] Zhang et al. (2011,July.). “Hypoxia-Inducible Factor Directs POMC Gene to Mediate Hypothalamic Glucose Sensing and Energy Balance Regulation.”PLOS BIOLOGY.[online]. 9(7), pp. 88. Available: [Feb. 26, 2012] [5] Rhiannan Williams et al. (2008, Aug.). “Adaptive sugar sensors in hypothalamic feeding circuits.” The scripps research institute. [online]. 105 (33). pp(126) Available: [Feb. 26, 2012] [6] Ambrose Dunn – Mynell et al. (2009, May.). “Relationship among Brain and Blood Glucose Levels and Spontaneous and Glucoprivic Feeding.”The Journal of Neuroscience.[online]. 29 (21). pp. 7015-7022 Available: [Feb. 26, 2012] [7] Annika Enejder et al. (2005, June.). “Raman spectroscopy for noninvasive glucose Measurements.”Bio medical Optics.[online]. 10 (3), pp . Available:[Feb. 26, 2012] [8] Douglas Stuart et al. (2006, Oct.). “In Vivo Glucose Measurement bySurface-Enhanced Raman Spectroscopy.”Analytical Chemistry.[online]. 78 (20). Pp. 55 Available: [Feb. 26, 2012] [9] Edward Stark. “Non invasive Glucose measurement and apparatus.” United States of America. 5433197, Jul. 18, 1995. [10] “Saline (medicine).” Internet:http://en.wikipedia.org/wiki/Saline_(medicine) [Feb. 4, 2012] [11] M. Silverstein. Spectrometric Identification of Organic Compounds. /ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /CropGrayImages true /GrayImageMinResolution 200 /GrayImageMinResolutionPolicy /OK /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageMinDownsampleDepth 2 /GrayImageDownsampleThreshold 2.00333 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /CropMonoImages true /MonoImageMinResolution 400 /MonoImageMinResolutionPolicy /OK /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.00167 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /CheckCompliance [ /None ] /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputConditionIdentifier () /PDFXOutputCondition () /PDFXRegistryName () /PDFXTrapped /False /CreateJDFFile false /Description > /Namespace [ (Adobe) (Common) (1.0) ] /OtherNamespaces [ > /FormElements false /GenerateStructure false /IncludeBookmarks false /IncludeHyperlinks false /IncludeInteractive false /IncludeLayers false /IncludeProfiles true /MultimediaHandling /UseObjectSettings /Namespace [ (Adobe) (CreativeSuite) (2.0) ] /PDFXOutputIntentProfileSelector /NA /PreserveEditing false /UntaggedCMYKHandling /UseDocumentProfile /UntaggedRGBHandling /UseDocumentProfile /UseDocumentBleed false >> ] >> setdistillerparams > setpagedevice