non-invasive glucose sensors

7/29/2019 Non-Invasive Glucose Sensors

1/12

41Losoya-Leal A y cols. State of the art and new perspectives in non-invasive glucose sensors

www.medigraphic.org.mx

ARTCULO DE REVISINVol. XXXIII, Nm. 1

J unio 2012pp 41 - 52

ABSTRAC T

Self-monitoring of blood glucose levels provides diabetic patients witha prompt method of measuring their blood glucose concentration, asopposed to conventional laboratory measurements. Frequent testing

aids patients in the prevention and detection of hyper or hypoglycemiaevents. Only invasive and minimally invasive glucose sensors are cur-rently commercially available. However, during the last couple of de-cades, work towards the development of a non-invasive glucose moni-tor has increased significantly among research groups with motivatingresults. Many techniques have been studied and implemented, eachwith their particular advantages and challenges. This paper presentsa qualitative review of different technologies of non-invasive glucosesensors: spectroscopy-based methods, transdermal extraction-basedmethods, fluorescence, electromagnetic variations and polarimetry.This study identifies strengths and opportunities of currently availableglucose monitoring techniques, as well as main characteristics andperformance variables for an ideal non-invasive monitor. The mostpromising approaches towards the development of a truly non-invasiveand clinically accurate glucose sensor are discussed.

Key words : Diabetes mellitus, non-invasive glucose monitoring, spec-troscopy, transdermal glucose extraction, fluorescence, polarimetry,electromagnetic variations.

RESUM EN

El automonitoreo de los niveles de glucosa en la sangre proporcionaa los pacientes diabticos un mtodo rpido para medir sus nivelesde glucosa en comparacin con los mtodos convencionales delaboratorio. Las mediciones frecuentes permiten a los pacientesprevenir y detectar episodios de hper o hipoglucemia. Actualmente,slo se encuentran disponibles comercialmente sensores invasivos omnimamente invasivos. Sin embargo, a lo largo de las ltimas d-

cadas ha aumentado significativamente el trabajo de los grupos deinvestigac in hac ia el desarrollo de un sensor de glucosa no invasivo,obteniendo resultados motivadores. Una gran cantidad de tcnicashan sido estudiadas e implementadas, cada una con ventajas yretos particulares. Este trabajo presenta una revisin cualitativa delos diferentes mtodos no invasivos de sensado de glucosa: mto-dos basados en espectroscopia, mtodos basados en la extraccintransdrmica, fluorescencia, variaciones electromagnticas y pola-rimetra. Mediante este estudio se identifican las principales ventajasy retos de cada tcnica, as como las principales caractersticas yvariables de desempeo para un sensor no invasivo ideal. Se discuten

State of the art and new perspec tives innon-invasive g lucose sensors

Losoya-Leal A,*Camacho-Len S,*

Dieck-Assad G,*Martnez-Chapa SO*

* Department of Electrical and ComputerEngineering, ITESM, Campus Monterrey.

Correspondence:Losoya-Leal AAv. Eugenio Garza Sada 2501, Monterrey, N.L.64849, Mxico.

Tel. 83582000, ext. 5414E-mail: [email protected]

Art c u lo rec ib ido: 12/d ic iem bre/2011A rt cu lo a ce p tad o : 14 /ma yo /2012

SOMIB

REVISTA MEXICANA DE

INGENIERA BIOMDICA

Este artculo puede ser consultado en ver-sin completa en: http://www.medigraphic.com/ingenieriabiomedica



2/12

42 Revista Mexica na d e Ingeniera Biomd ica volumen XXXIII nmero 1 Junio 2012


INTRODUCTION

Self-monitoring of blood glucose levels provides dia-betes patients with a continuous, robust and reliableprocedure to determine blood glucose concentra-tion, as opposed to conventional laboratory tests.Frequent verification of glucose levels is crucial inthe treatment of diabetes, and it aids patients inpreventing and detecting hyper or hypoglycemiaevents. Many commercial glucose monitors exist,

most of them requiring a small drop of blood ob-tained through a puncture of the skin with a lancet,usually on the tip of the fingers (commonly referredto as finger-stick testing). These are invasive glucosemonitors, and are highly uncomfortable for patientsdue to the frequent puncturing of up to 10 times aday. Additionally, measurements with these devicesusually present an error of approximately 6-7%1.

This percentage may increase depending on thesize and quality of the sample, human error duringsample extraction, faulty calibration, humidity, andlack of hygiene in the extraction area2.

These problems have generated great interest

in the exploration of new non-invasive monitoringtechniques. The aforementioned exploration shouldlead to systems that will be able to detect andestimate glucose concentrations by meticulouslyeliminating different sources of interference. Oneof the main goals of these systems is to achievehigh precision in their measurements, providing atleast the same measurement errors as conventional,finger-stick methods. Unfortunately, none of thefew non-invasive self-monitoring devices that wereapproved over the years by the US Food and DrugAdministration (and consequently introduced to themarket) offered low measurement error to replace

lancet approaches, leading to their eventual dis-appearance.

Currently, only invasive and minimally invasive glu-cose sensors are commercially available. However,during the last couple of decades, work towards thedevelopment of a non-invasive glucose monitorhas increased significantly among research groupswith promising results. Many techniques have beenstudied and implemented, generating particularbenefits and drawbacks.

This paper presents a qualitative review of non-invasive glucose sensors, which includes operatingprinciples as well as the strengths and weaknessesof each sensor type. The overview of glucosemonitoring presents three main categories: inva-sive, minimally invasive and non-invasive glucosemonitoring. The most representative non-invasiveglucose sensing techniques are described. Fi-nally, the most important specifications of idealnon-invasive glucose sensors are summarized

over a table which identifies the most promisingapproaches.

1) GLUCOSE MONITORING

Blood glucose monitors can be categorized inthree well-defined groups based on their interac-tion with a patients body. The first group includesinvasive monitors, most based on finger-stick test-ing; a procedure where a small blood sample isextracted by puncturing of the skin with a lancet.

The extracted sample is deposited over a chemi-cally active disposable test strip that is inserted to

the monitor, which displays the glucose concentra-tion. The device is based upon an electrochemicalreaction between the blood and chemical analytesfrom the strip. Other invasive glucose measurementdevices include hospital bedside monitors, whichare attached to the patients intravenous lines toobtain blood samples first, and provide glucoseestimations afterwards by optical methods such asnear-infrared spectrometry.

The second group consists of minimally invasivedevices, which rely on the measurement of intersti-tial fluid (IF) by means of a small sub-dermal implant.

This implant converts glucose concentration from

the patients IF into an electronic signal3. Otherminimally invasive monitors involve microdialysistechniques applied in the extraction of IF samplesfora p oster ior i analysis. Microdialysis is a very impor-tant tool for sampling free drug concentrations fromany tissue to determine their pharmacokinetics. Incombination with a suitable detection technique,microdialysis allows monitoring of time-dependentchanges in tissues chemistry. In microdialysis forglucose measurement, the extraction occurs by

las tcnicas ms prometedoras hacia el desarrollo de un sensor conprecisin c lnica y verdaderamente no invasivo.

Palabras clave: Diabetes mellitus, monitoreo no invasivo de glucosa,espectroscopia, extraccin transdrmica de glucosa, fluorescencia,

po larimetra, va riac iones elec tromagntic as.


3/12



means of a fiber generally inserted in the adiposetissue of the abdominal region4.

The third group represents all apparatuses usingnon-invasive sensors. Non-invasive glucose sensingresearch strives to provide an effective solution to

the discomfort suffered by the perforation of the skinthat many invasive devices require. Other advan-tages that make this approach attractive are thepossibility of real-time and continuous monitoringas well as the elimination of infection risks. Figure 1summarizes the three main non-invasive glucosemonitor groups.

The reliability of a glucose monitor is directlyrelated to its accuracy. The technical accuracyof a blood glucose meter can be defined as thecloseness of agreement between the measuredquantity value and a reference quantity of glucose.

This reference quantity is usually obtained from alaboratory test.Standards organizations and professional soc i-

eties differ on accuracy acceptability criteria. TheAmerican Diabetes Association (ADA) has recom-mended that glucose meters agree to within 15%of the laboratory method at all concentrations, witha future performance goal of 5% agreement at

all glucose concentrations. Likewise, the FDA andthe International Organization for Standardization(ISO) have set accuracy criteria to 20 mg/dl forlevels below 100 mg/dl or 20% for glucose levelshigher than 100 mg/dl for at least 95% of results5.

Given that no single standard exists, the approvalof a meters accuracy will vary depending of thecountry and/or the standards organization beingfollowed.

2 ) NO N- INVASIVE GLUC O SE SENSORS

Apparatuses and devices to measure glucose us-ing non-invasive techniques are classified in threegroups: spectroscopic techniques, transdermal glu-cose extraction and other techniques. Non-invasiveglucose sensors have been studied and developed

by research teams around the world. The followingsections discuss in some detail the different tech-niques explored.

Spectroscopic techniques

Spectroscopy is the study of objects based on theirwavelength spectrum, both when they emit or ab-

F i g u r e 1 . Over-view of non-in-vasive glucosemonitoring tech-niques.

. ed r hi . . x


4/12



sorb light. Furthermore, through the spectroscopictechnique of spectrophotometry, the concentrationof a given chemical species can be assessed.Figure 2 shows the basic components of a spec-trophotometry system. White light from a source(containing a wide range of wavelengths) is directedtowards an entrance slit and the collimated lightpasses through a diffraction system (e.g. a prism ora diffraction grating). The diffractive element divides

the incoming light into its wavelength components,which are then collimated by an exit slit selectinga particular wavelength range. The selected rangeis directed towards a sample and the transmittedlight is finally directed towards a detector system.

Four spectroscopic techniques that have beenapplied to non-invasive glucose monitoring are de-scribed next: near-infrared absorption spectroscopy,Raman spectroscopy, bioimpedance spectroscopyand thermal emission spectroscopy.

Near-infrared absorption spectroscopy

Infrared spectroscopy is a well established andconstantly developing analytical technique whichallows for the rapid, high-throughput, non-destruc-tive analysis of a wide range of sample types6. Thistechnique is based upon the principle that everymolecule has resonance absorption peaks whichare directly related to the molecules concentrationin a sample7. Because of this, radiation may be di-rected through body tissue and the exiting light canbe analyzed to determine the content of glucose.

Near-Infrared (NIR) absorption spectroscopy tech-niques have long been mainstays of nondestruc-tive chemometric analysis and therefore, they holdgreat potential for the development of non-invasiveblood glucose measurement techniques8.

The main obstacles for in vivo blood glucosemeasurement by NIR absorption spectroscopy arethe interference from some absorbing species;multiple scattering in skin, muscle and bone; and

the strong absorption bands of water9

.Different wavelength ranges have been utilizedby research groups based on the target bodyarea (e.g. eye and finger or forearm skin), most ofthem within the therapeutics wavelength windowof 600-2,500 nm. Wavelengths within this range areselected due to the weak absorbance by water,the high energy of the signal source, as well as theavailability of commercial light transducers1.

A novel technique, named pulse glucometry9,aims to remove the aforementioned interferencesby obtaining data from a blood-only sample us-ing instantaneous differential NIR spectrometry (at

the wavelength range from 900 to 1,700 nm.). Thistechnique is based on two almost instantaneousmeasurements over a tissue sample which is usu-ally the tip of a finger. Thus, changes in opticalabsorption between both measurements dependof a blood volume change produced by cardiacpulses. By a subtraction process, the interference ofbasal elements is then removed. Pulse glucometryhas been tested in humans obtaining promisingresults9.

Figu re 2 . Basic components of a spec-trophotometer.Light source

Entrance slit

Diffraction system

Exit slit

Sample

Detector system


5/12



Another technique used to remove interferencecaused by strong absorption bands of water iscalled attenuated total reflectance (ATR), in whicha diamond prism is placed in contact with the fingertip. When an infrared light beam travels through a

prism, it reflects several times, forming an evanes-cent wave which extends into the sample. The beamis then collected by a detector as it exits the crystal.Changes in the light absorption are related to theconcentration of blood glucose10.

NIR spectroscopy data relates to many overlap-ping bands and thus requires multivariate analy-sis6. Classical least squares method8, partial leastsquares11, support vector machines12 and artificialneural networks13, have been utilized as multivariatecalibration models for in vivospectroscopic analysis.

A potential non-invasive glucose sensor based on

NIR spectroscopy requires the miniaturization of aspectrophotometers components, so as to providea portable device which satisfies patients need offrequent testing. Work towards the miniaturizationof a spectrophotometers components for non-invasive sensing of glucose has been presented14,15.

Raman spectroscopy

When radiation impacts upon a sample, most ofthe incident light suffers Rayleigh scattering and asmall portion of light undergoes frequency shifts8.

The measurement of those shifts is known as Raman

spectroscopy6

. Water has a poor scattering response,thus, Raman spectroscopy provides a spectralsignature that is less influenced by water than NIRspectroscopy8. Therefore, measurements can bemade directly from bio-fluids, although, the effectis weak, making the collection times relatively long6.

To avoid long collection times, high power lasersare an alternative; however, these are harmful forhuman tissues. This has led to efforts in developingcrystal structures so that the Raman scattering maybe initiated and enhanced for a low power incidentbeam. The use of noble metal clusters inside a pho-tonic crystal structure was proposed. This approach

would result in surface-enhanced Raman scatter-ing, providing enhancements of up to one milliontimes16. Monochromatic lasers are the tendencyregarding the sources utilized when applying thistechnique. The most common wavelengths rangesare from 785 nm17,18 to 830 nm17,19,20.

Studies show that Raman spectroscopy is feasiblein ex vivo porcine eyes17 and transcutaneously inhuman forearms19. A portable device has also beendescribed20, but proof of its functionality as a non-

invasive glucose monitor in humans has not beenpresented so far.

Bioimpedance spectroscopy

Bioelectrical impedance analysis (BIA) is a diagnostictechnique based on the study of the passive electri-cal properties of biological tissues. The practical useof these electrical passive properties initiated nearthe middle of the twentieth century. Different tech-niques resulted in a collection of methods that arenow used for multiple applications: determination oftotal body water, fat free mass, tissue characteriza-tion, apnea monitoring, venous thrombus detection,tomography, cardiography, pneumography andblood compounds analysis21. Usually, these methodsshare three common advantages: use of low-cost

instrumentation, easy application in practice, andon-line monitoring availability.BIA uses electrodes to apply low intensity currents

in physiological fluids or tissues. The resulting voltagereflects changes in dielectric or dimensions of thetarget, therefore, enabling the monitoring of chemi-cal compositions or even physiological events inanimal and human bodies. Important features ofBIA include low cost, fast response, simple imple-mentation and safety. Such characteristics, mostlyassociated to the development of microelectronics,make BIA a promising tool for tissue charac teriza-tion and compound quantification, especially by

dielectric spectroscopy at high frequencies. Di-electric spectroscopy is a technique in which thedielectric properties of a medium are measured asa function of frequency22.

A wrist glucose monitor based on impedancespectroscopy has been presented23. This monitorsamples information from an LC resonance circuitusing the skin as a dielectric. An important disadvan-tage is that it requires a calibration process in whichthe patient must rest for at least 60 minutes beforestarting measurements. Even though the frequencyband was chosen in order to decrease the sensitivityto side effects, other environmental variables also

produce measurement errors. For example, sweatand relocation of the sensor produce inaccura-cies. Moreover, variations in temperature affect thesensors signal since they have an impact on thepermittivity of water and other molecules.

Th e r m a l e m i ssi o n sp e c t ro s c o p y

Thermal emission spectroscopy (TES) measures IRsignals generated in the human body as a result of


6/12



glucose concentration changes. This technique isbased on the principle that natural mid-IR emissionfrom the human body, especially from the tym-panic membrane, is modulated by the state of theemitting tissue. Its selectivity is based on the same

principle as the selectivity of the absorption spec-troscopy technique for analyte measurements24.

Plancks law describes a relationship between theradiant intensity, spectral distribution, and tempera-ture of a black body. The human body is an excel-lent black body emitter of mid-IR light. The spectralcharacteristic of thermal emission is influencedby the individuals tissue composition and analyteconcentrations. On the other hand, Kirchhoffs lawstates that, for the same wavelength, absorbanceequals monochromatic emissivity when the entirebody is at the same temperature. Since the tym-

panic membrane shares its blood supply with thehypothalamus, which is the bodys temperatureregulation center, it is an ideal site to measure bodytemperature.

A tympanic thermometer measures the integralintensity of IR thermal radiation. A sensor insertedin the ear canal can obtain a clear view of themembrane and its blood vessels to measure theemitted IR radiation. When compared with the theo-retical black body radiation described by Plancksand Kirchhoffs laws, the membranes IR radiationis spec trally modified by tissue composition. Theinstrument optically receives IR radiation from the

target object, such as the tympanic membrane. Thedetecting system consists of an optical IR filter setand a thermopile detector sensitive to an IR region.One of the sensing elements is covered by an IRfilter sensitive to the IR glucose signature, while anappropriate filter that does not have spectral bandscharacteristic to the measured analyte covers theother sensing area. Spectrally modified IR radiationfrom the tympanic membrane illuminates both win-dows. The difference between the intensities of thetwo radiation paths provides a measure proportionalto the analyte concentration25.

A non-invasive prototype based on thermal emis-

sion in the mid-IR spectral region has shown glucosemeasurements with acceptable accuracy. One ofthe advantages of this system is that individual dailycalibrations are not required24. One of the draw-backs is that the intensity of the infrared radiationemitted by an eardrum is affected not only by itstemperature but also by its thickness. Although theclinical results obtained using TES are promising, theyare not yet considered acceptable under clinicalaccuracy standards.

Tr a n s d e r m a l g l u c o se e x t ra c t i o n

Measuring glucose is also possible if the substanceis extracted from interstitial fluid and the analysis isperformed on an ex pos t fac to basis. Three differ-

ent techniques are presented in technical literature:reverse iontophoresis, sonophoresis and dead skinthermal ablation.

Reverse iontophoresis

Iontophoresis is a well known and studied techniquein which charged and uncharged polar compoundsare driven across the skin by the application of lowDC currents26. Iontophoresis has been widely usedfor drug delivery purposes in dermatology, ophthal-mology, otolaryngology among other disciplines27.

Its counterpart, reverse iontophoresis, is the extrac-tion of these compounds through the skin by thesame concept of low electric current application.

In reverse iontophoresis a low level DC currentis conducted through the skin between two elec-trodes. The migration of ions creates an electro-osmotic solvent flow toward the negative electrode,which makes the transport of polar unchargedspecies such as glucose possible. Skin has a nega-tive charge, thus, it is permselective to cations. Theaforementioned flow is consequently used for theextraction of the compounds via the skin28.

The monitoring of two interacting biomolecules,

glucose and lactate, has been demonstrated tohave particular success with this technique29. In thecase of blood glucose monitoring, one device byCygnus Inc. gained FDA approval in 2001 and wasmade available commercially. The GlucoWatchBiographer made non-invasive glucose measure-ments through the skin every 10 minutes for a periodof 13 hours30. However, the GlucoWatch presentedvarious limitations. First of all, the FDA suggestedthat the device should not be used as a substituteor replacement of finger-stick measurements, butas a support of standard home glucose monitoringdevices. For the determination of insulin doses, the

FDA also suggested confirmation of the GlucoW-atchs measurement with a traditional finger-stickmeasurement. These suggestions automaticallycompromised its purpose. Other drawbacks werethat device presented a 15 minute lag comparedwith finger-stick readings, and required 2 hours ofwarm-up and calibration, which could result in er-roneous readings if not done correctly. GlucoWatchfailed to satisfy patients needs and expectationsdue to these and other inconveniences such as


7/12



inconsistency in measurements, frequent skin irrita-tion, itching and blistering, intermittent shutting off,and skipping of readings due to movements. Con-sequently, this device failed to offer reliable mea-surements and is no longer commercially available.

Areas of further investigation regarding reverseiontophoresis are the effects of environmentalvariables in measurements (e.g. temperature, pH),adverse effects due to skin polarization (e.g. skinburning and blistering) and other factors such asperspiration and skin resistance31. Additionally, theuse of the sodium ion as an internal standard hasbeen reported to be of assistance in the determi-nation of glycemic events by reverse iontophoresiswithout the need of calibration with a blood sample.

Sonophoresis

Sonophoresis is a technique that increases cutane-ous permittivity to interstitial fluid by micro-vibrationsproduced when ultrasound waves are directed tothe tissue with a piezoelectric transducer. This tech-nique is similar to reverse iontophoresis and it usuallyenhances transdermal delivery of drugs.

A technique that implements a combination ofa low-profile cymbal array and an electrochemicalglucose sensor has been developed32. The sensorconsists of amperometric electrodes combined witha novel glucose oxidase hydrogel. Glucose con-centrations can be determined by increasing cu-

taneous permittivity to interstitial fluid, enabling thetransportation of glucose to the epidermis surface.Some experiments show that through the placementof the aforementioned array in the zone of exposureto the ultrasound waves, glucose concentrationmay be determined, but error ranges during testingwere unacceptably high. After comparing measure-ments with commercial glucose meters readings,the difference between values of each respectivetechnique was around 90 mg/dl. Results for testingin humans have not been presented yet.

Dead skin thermal ablation

Glucose molecules are unable to reach the skinsurface because of their inability to diffuse acrossthe layers of dead skin cells. However, by openinga microscopic region of this layer through thermalablation, glucose is allowed to diffuse naturally tothe skin surface33. Thermal ablation has successfullybeen used for transdermal drug delivery and thepotential of this technique for vaccination processeshas been recognized34.

A dermal patch with micro-heater elements hasbeen fabricated and tested on surrogate humanskin with promising results33. This patch is able tosample glucose molecules from interstitial fluidin a non-intrusive manner by applying a highly

localized heat pulse to the dead-skin layer untilablation is reached. Due to the strict control ofthe applied heat pulse, only the dead-skin layer isablated, leaving the underlying tissue and nervesunaffected. Therefore, this technique is consideredas non-invasive.

O t h e r t e c h n i q u e s

In addition to Spectroscopy and Transdermal Glu-cose Extraction, three other techniques which canperform the measurement of glucose concentration

are reported in technical literature: Fluorescence,Electromagnetic Field Variations and Polarimetry.

Fluorescence

Fluorescence-based sensors constitute a growingclass of glucose sensors with significant contributionsto the field over the last few years. This techniquehas been encouraged by special opportunitiespresented by the fluorescence principle for biologi-cal analysis such as fast, noninvasive, very highlysensitive and in some cases, reagent-less and non-consumable sensing35,36. Different approaches

have been explored in fluorescence-based sensors,including: auto-fluorescence, tattoo-based andtear-monitoring sensing.

One of the techniques for fluorescence-basednon-invasive glucose monitoring is based on themeasurement of cell auto-fluorescence due toNAD(P)H and signaling of changes in extracellularglucose concentration by fluorescent markers ofmitochondrial metabolism37. However, these ex-periments have been performed using in vitro cellculture models, thus requiring further investigationfor monitoring in humans.

A nanosensor similar to a tattoo dye consisting

of 120 nm-diameter hydrophobic spheres is underdevelopment38. The spheres consist of a glucosedetecting molecule, a color-changing dye and aglucose-mimicking molecule. The detecting mol-ecule will bind with either the glucose moleculesor the mimicking molecule, depending on theconcentrations of glucose in the medium. Conse-quently, the tattoo will fluoresce differently underan IR source according to the concentration ofglucose. The sampling process repeats itself every


8/12



few milliseconds with a 20 minute lag respect toactual blood glucose levels. Results have beenpositive on mice but the technique is yet to betested in humans.

Another approach is based on the determination

and monitoring of tear glucose levels, which areknown to track glucose concentrations in blood39.

The technique relies on disposable and color-less contact lenses embedded with boronic acidprobes (containing fluorophores), which are knownto chelate monosaccharides and thus provide aglucose-sensing mechanism through the lens colorvariations40. These probes are compatible with theinternal pH and polarity of the lenses. The extent ofchelation can be monitored using different tech-niques, for example: fluorescence, colorimetry andpolarization41. The reversible nature of this technique,

as well as its sensitivity, suggests its feasibility as acontinuous non-invasive monitor of physiologicalglucose.

Electromagnetic field variations

A change in blood glucose concentration causes avariation in the bloods dielectric parameters suchas its conductivity. An electromagnetic sensor basedon Eddy currents due to changes in the electro-magnetic field can determine the concentrationof glucose through the aforementioned variations42.

The principle of the sensor is based on two coils,

with a separation of 40 mm between them. Theconductivity fluctuations as a result of the changein glucose concentration involve the variation ofthe coilsoutput signal. The technique consists onapplying a signal with a frequency of approximately4 MHz to the primary coil, while the output signal ismeasured on the secondary coil43.

Polarimetry

The use of polarimetry in the detection of analyteconcentrations has existed for several years. Thetechnique relies on the ability of optically active

analytes to rotate the plane of polarization of inci-dent linearly polarized light44. One possible analytefor this application is glucose, which in the aqueoushumor of the eye closely mimics glucose levels inblood45. Aqueous humor is preferred as a target forthis technique, as it does not present the high scat-tering coefficients that skin does. Additionally, thistarget is convenient due to the short delay time ofonly a few minutes with respect to blood glucoselevels46.

Parameters involved in the determination of ananalytes concentration by polarimetry are thespecific rotation of the light source, the samplepathlength and the observed rotation of linearlypolarized light caused by the analyte47. Glucose

concentration will be directly proportional to theobserved rotation in polarization for a known path-length inside the sample.

Although polarimeters have existed for severalyears, it has not been until recently that their sen-sitivity has increased enough to be applied in themeasurement of the low glucose concentrationsseen in physiological conditions. The technique haspresented obstacles and drawbacks, such as theinability to provide enough system stability to ac-curately measure the detected signal consistentlyover large periods of time, as well as the limits im-

posed by the complicated shape of the cornea

44,45

.Furthermore, humor glucose levels are lower thanblood levels and there are many biomolecules in itthat contribute with confounding optical rotations47.

Several research groups have studied differentapproaches to overcome polarimetrys drawbacksin glucose measurements. Two different approacheshave been presented in literature45, as well as therespective geometrical analysis for optically ac-cessing the aqueous humor of the human eye. Adigital closed-loop control system that improves thesystems repeatability and stability without sacrific-ing accuracy has been utilized as well44. Another

approach using a true phase measurement tech-nique has been shown to be theoretically moreimmune than others to potentially confoundingeffects such as corneal birefringence and rotation,reducing additive amplitude noise effects47. How-ever, a full blown version of a polarimetric deviceapplied to glucose monitoring in humans has notbeen presented so far.

Although many ongoing research projects havereported developments of truly non-invasive bloodglucose measurement techniques, more workand innovation efforts are still necessary towardsthe design of a portable and clinically accurate

device.

3 ) SUM M ARY OF STRENG THS AND O PPO RTUNITIESIN GLUCOSE MONITORING

Tables 1-3 summarize the most important advan-tages and challenges of each of the techniquesdiscussed in this paper, divided by their respectivegroups: spectroscopic techniques, transdermalglucose extraction and other techniques.


9/12



Ta b l e 1 . Advantages and challenges in non-invasive glucose monitoring for spec troscopic techniques.

Advantages Challenges Performance parameters

NIR spectroscopy Well established analytical Requires multivariate Average absolute error:

6-13 technique analysis 1.1 mmol/l Non-destructive analysis Device exists in the Root mean standard error: Efficient removal of most macroscale only, 1.02-1.88 mmol/l

interferents requires miniaturization Mean prediction error: Successfully tested in humans Scattering in skin 9.3%

Raman spectroscopy Spectral signature less Long collection times Concentration estimated6,16-20 influenced by water compared Collection times reducible down to: 0.7 mmol/l

to NIR spectroscopy by with lasers, but not (in water) Measurements directly from without harm 7.2 mmol/l (in

biofluids Proof of functionality in bioreactor material) Feasible transcutaneously in humans not presented 2.5 mmol/l (in serum

forearms and plasma samples)

Bioimpedance Low cost instrumentation Very long calibration Correlation coefficientspectroscopy Simple application in practice process with current blood glucose in low21-23 On-line monitoring available devices control outpatient

Safe and with fast response Susceptible to interference conditions:by different phenomena 0.64such as movement, sweatand temperature

Thermal emission Prototype has shown Intensity of radiation by Mean error: 0.638spectroscopy acceptable accuracy eardrum is affected by its mmol/l24-25 No daily calibrations needed thickness, which is variable

in humans Results of measurements

not acceptable under

clinical standards

4) CONCLUSIONS

By means of this review, many desirable charac-teristics and performance variables were identifiedas key elements in the development of a truly non-invasive and clinically accurate glucose sensor.

These characteristics include:

Clinical accuracy for reliable results compa-rable to invasive and minimally-invasive tech-

niques. Portability for fast transportation and availability. Short measurement time in order to reduce

artifacts due to physical manipulation of thedevice (e.g. patient movement during themeasurement, etc.) as well as for comfortable-ness of the patient.

Simplicity in its calibration, in order to avoid hu-man errors in this task that may lead to faultyresults.

Short or no lag at all regarding glucose levelscompared with finger-stick or other invasive andminimally invasive techniques, in order to obtainreal-time glucose concentrations.

Robustness in its design to minimize the effects ofpossible interference sources in order to provideconsistency in measurements. These interfer-ence sources include transpiration, movement,temperature, pH, scattering by body tissues andphysiological fluids.

High sensitivity and selectiveness in com-parison with invasive and minimally-invasivetechniques to provide significant results forthe patients glucose control schemes. Sen-sitivity improvements may be achieved bythe parallel monitoring of more than oneparameter. Selectiveness is important (e.gin spectroscopic techniques) given that thecollected spectroscopic information mustcontain a selective signature for glucose


10/12



Este documento es elaborado por Medigraphic

Ta b le 2 . Advantages and challenges in non-invasive glucose monitoring for transdermal glucose extraction techniques.


Reverse Work is in advanced stages as Devices have functioned Median correlation

iontophoresis devices have been developed only as auxiliary to invasive coefficient with blood26-31 and tested in the market measurements glucose: 0.87 There is lag and

inconsistency in comparisonwith finger-stick method

High calibration times Not unharmful, as devices

produce irritation andblistering in patients

Affected by environmentalvariables

Sonophoresis Preliminary in-vivo results Experimental error ranges (Testing in rats)32-33 have been described unacceptably high Correlation with blood

predicting glucose levels in rats Results of testing in glucose of: 0.93humans not reported Mean relative error: 15%

Dead skin ablation Tested in humans with Fabrication process Not available33-34 favorable results presents complications

Not painful for patients as only that compromise thedead skin is affected quality of the sensors

Still in prototypical stages

Ta b l e 3 . Advantages and challenges in non-invasive glucose monitoring for other techniques.


Fluorescence Different approaches have Has not been tested in Not available35-40 proven the viability of the humans, only on in-vitro

technique cell culture models Painless and reversible

Electromagnetic Linear relationship between Results are based on Glycemic Sensitivityfield variations the modulus of output/input in-vitro experiments (for blood inside a plastic42-43 signal and glucose There is still no testing tube): 4.4 mmol/l

concentrations in humans Reproducibility in Work is still in preliminary

measurements stages and no integrateddevice has been presented

Interference by diversephenomena has not beenaddressed

Polarimetry Aqueous humor does not Low system stability and Resolution:44-47 present the high scattering consistency in measurements 0.55 mmol/l

skin does over large periods of time Aqueous humor presents a Shape of the cornea

short time delay of only a few imposes limitsminutes respect to blood Humor glucose levels areglucose levels lower, and there are more

There are advances in the confounding biomoleculesreduction of confounding involvedeffects and in the stability and No successful device hasrepeatability of measurements been presented


11/12



relative to a ll the other components that canimpact the signal.

No side-effects to the patient such as irritation,burning or blistering to provide a truly painlessmeasurement, thus improving the quality of life

of diabetes patients. Reasonable instrumentation cost to provide a

cost-effective and marketable solution.The spectroscopic techniques described seem

highly attractive due to the non-requirement of sen-sor implantation orex-vivoblood sampling. The mainobstacles of near-infrared absorption spectroscopyinclude confounding by other optical absorbersand strong absorption bands of water. The PulseGlucometry technique aims to remove such inter-ferences and has proven to be clinically accuratein the macro-scale; therefore, it represents one of

the most promising approaches to truly non-invasivemonitoring.Other techniques described in this review also

possess distinct technical advantages and favor-able conditions for their successful implementation,and are dependent on their further refinement. Bio-impedance spectroscopy has also shown viability,simplicity, low cost and favorable results, and onlyrequires work towards a more robust device regard-ing sources of inaccuracies. Additionally, glucoselevels measurement by transdermal extraction ofthe interstitial fluid has previously reached a com-mercial level and thus only requires refinement

of specific issues such as calibration. It has beenreported that use of the sodium ion as an internalstandard could refine the determination of glycemiaby reverse iontophoresis without requiring calibrationwith a blood sample.

Evolution in sensor technology and microelectron-ics is bound to open the door to new approachesand methods that should aid the refinement of thediscussed techniques, leading to truly non-invasiveglucose sensors.

REFERENC ES

1. Ferrante AC, Wolf B. Current development in non-invasiveglucose monitoring. Medical Engineering & Physics 2008;30: 541-549.

2. Vesper HW, Myers GL. Approaches for improving glucosemonitor measurements for self-monitoring of blood glucose:from measurement harmonization to external quality assess-ment programs. Journal of Diabetes Science and Technology2007; 1(2): 2964-2967.

3. Thennadil SN, Rennert JL, Wenzel BJ , Hazen KH, Ruchti TL, BlockMB. Comparison of glucose concentration in interstitial fluid,and capillary and venous blood during rapid changes inblood glucose levels. Diabetes Technology & Therapeutics,Sept. 2001.

4. Krogstad AL, Jansson PA, Gissln P, Lnnroth P. Microdialysismethodology for the measurement of dermal interstitialfluid in humans. British Journal of Dermatology J une 1996;134(6):1005-1012.

5. Tonyushkina K, Nichols J. Glucose meters: a review of thetechnical challenges to obtaining accurate results. J ournal

of Diabetes Science and Technology J uly 2009; 3(4): 971-980.

6. Ellis DI, Goodacre R. Metabolic fingerprinting in diseasediagnosis: biomedical applications of infrared and Ramanspectroscopy. Analyst 2006; 131: 875-885.

7. Mendelson Y, Clermont AC, Peura RA, Lin B. Blood glucosemeasurement by multiple attenuated total reflection andinfrared absorption spectroscopy. IEEE Transaction on Bio-medical Engineering 1990; 37: 458-465.

8. Biomedical Photonics Handbook. 1st ed., Duke University,Durham, NC, 2003.

9. Yamakoshi Y, Ogawa M, Yamakoshi T, Satoh M, Nogawa M,Tanaka S, Tamura T, Rolfe P, Yamakoshi K. A new non-invasivemethod for measuring blood glucose using instantaneousdifferential near infrared spectrophotometry. 29th Annual In-ternational Conference of the IEEE, Lyon, France Aug, 2007:

2964-2967.10. Ishizawa H, Muro A, Takano T, Honda K, Kanai H. Non-invasive

blood glucose measurement based on ATR infrared spec-troscopy. SICE Annual Conference 2008: 321-324.

11. Tenhunen J , KopolaH, Myllyl R. Non-invasive glucosemeasurement based on selective near infrared absorption;requirements on instrumentation and spectral range. Mea-surement 1998; 24: 173-177.

12. Ogawa M, Yamakoshi Y, Satoh M, Nogawa M, YamakoshiT, Tanaka S, Rolfe P, Tamura T, Yamakoshi K. Support vectormachines as multivariate calibration model for predictionof blood glucose concentration using a new non-invasiveoptical method named pulse glucometry. Proceedings ofthe 29th Annual International Conference of the IEEE EMBS2007; 29: 4561-4563.

13. Ham FM, Cohen GM, Cho B. Glucose sensing using infrared

absorption spectroscopy and a hybrid artificial neural net-work. Annual International Conference of the IEEE Engineeringin Medicine and Biology Society 1991; 13: 1574-1576.

14. Losoya-Leal A, Camacho-Leon S, Martinez-Chapa S. Designof a CMOS-MEMS diffraction grating for near-infrared spec-troscopy of biomolecules. Proceeding of the 7th IEEE Confer-ence on Electronics, Robotics and Automotive Mechanics(CERMA), Cuernavaca, Mexico, Oct. 2010: 737-740.

15. Martinez-Torteya A, Dieck-Assad G, Martinez-Chapa S, Ca-macho-Leon S. Modeling and simulation of a CMOS-MEMSinfrared thermopile; Simulation: Transactions of the Societyfor Modeling and Simulation International, England, 2011 (inreview).

16. Handapangoda CC, Premaratne M, Paganin DM. Simula-tion of a device concept for noninvasive sensing of bloodglucose levels. Third International Conference on Information

and Automation for Sustainability, 2007; 3: 31-34.17. Ergin A, Thomas GA. Non-invasive detection of glucose in

porcine eyes. Proceedings of the IEEE 31st Annual NortheastBioengineering Conference 2005; 31: 246-247.

18. Stewart DA, Yuen JM, Shah N, Lyandres O, Yonzon CR, Glucks-berg MR, Wals JT, Van Duyne RP. In vivo glucose measure-ment by surface-enhanced Raman spectroscopy. AnalyticalChemistry 2006; 78: 7211-7215.

19. Enejder AMK, Scecina TG, Oh J, Wei-Chuan MH, Sasic S,Horowitz GL, Feld MS. Raman spectroscopy for noninvasiveglucose measurements. J ournal of Biomedical Optics 2005;10: 1-10.


12/12



20. Ergin A, Vilaboy J, Tchouassi A, Greene R, Thomas GA. De-tection and analysis of glucose at metabolic concentrationusing Raman spectroscopy. Proceedings of the IEEE 29thAnnual Bioengineering Conference 2003; 29: 337-338.

21. J ossinet J, Trillaud C. Technical improvement of a dual fre-quency EIT tomography for bioelectrical tissue characteriza-

tion. Proceedings of the IEEE/EMBS 14th Annual InternationalConference 1992; 14: 1712.

22. Amaral C, Brischwein M, Wolf B. Multiparameter techniquesfor non-invasive measurement of blood glucose. Sensorsand Actuators 2009; 140: 12-16.

23. Caduff A, Hirt E, Feldman Y, Ali, Heinemann L. First humanexperiments with a novel non-invasive, non-optical continu-ous glucose monitoring system. Biosensors & Bioelectronics2003; 19: 209-217.

24. Buchert J M. Thermal emission sprectroscopy as a tool fornon-invasive blood glucose measurements. Proceedings ofthe SPIE, 2004; 5566: 100-111.

25. Malchoff CD, Shoukri K, Landau J I, Buchert J M. A novel non-invasive blood glucose monitor. Diabetes Care 2002; 25(12):2268-2275.

26. Sieg A, Guy RH, Delgado-Charro MB. Electroosmosis in

transdermal iontophoresis: implications for noninvasive andcalibration-free glucose monitoring. Biophysical J ournal2004; 87(5): 3344-3350.

27. Singh P, Maibach HI. Iontophoresis in drug delivery: basicprinciples and applications. Critical Reviews in TherapeuticDrug Carrier Systems 1994; 11: 161-213.

28. Singh P, Maibach HI. Iontophoresis: an alternative to the use ofcarriers in cutaneous drug delivery. Advanced Drug DeliveryReviews 1996; 18(3): 379-394.

29. Ching CTS, Buisson Y, Connolly P. The effect of pulsed bipolardc on the simultaneous extraction of glucose and lactateby reverse iontophoresis. Sensors and Actuators B: Chemical2008; 129(2): 504-509.

30. US Food and Drug Administration. Glucowatch G2 biographer-P990026/S0008, Device Approvals and Clearances, 2009.[Online]. Available: http://www.fda.gov/MedicalDevices/

ProductsandMedicalProcedures/DeviceApprovalsandClear-ances/Recently-ApprovedDevices/ucm083294.htm

31. Kim DW, Kim HS, Lee DH, Kim HC. Importance of skin resis-tance in the reverse iontophoresis-based noninvasive glu-cose monitoring system. Engineering in Medicine and BiologySoc iety. IEMBS 04. 26th Annual International Conference ofthe IEEE, 2004; 1: 2434-2437.

32. Kost J , Mitragotri S, Gabbay, Pishko M, Langer R. Transdermalmonitoring of glucose and other analytes using ultrasound.Nat Med 2000; 6: 347-350.

33. Parnjape M, Garra J , Brida S, Schneider T, White R, Currie J .A PDMS dermal patch for non-intrusive transdermal glucosesensing. Sensors and Actuators A 2003;104:195-204.

34. Coulman SA, Antsey A, Gateley C, Morrisey A, McLoughlin P,Allender C. Microneedle mediated delivery of nanoparticlesinto human skin. International J ournal of Pharmaceutics2009;366: 190-200.

35. Moschou EA, Sharma BV, Deo SK, Daunert S. Fluorescenceglucose detection: advances toward the ideal in vivo bio-

sensor. J ournal of Fluorescence 2004; 14(5): 535-547.36. Pickupa J C, Hussain F, Evans ND, Rolinski OJ , Birch DJ S.

Fluorescence-based glucose sensors. Biosensors and Bio-electronics 2005; 20: 2555-2565.

37. Evans NC, Gnudi L, Rolinski OJ , Birch DJS, Pickup JC. Non-invasive glucose monitoring by NAD(P)H autofluorescencespectroscopy in fibroblasts and adipocytes: A model for skinglucose sensing. Diabetes Technology & Therapeutics 2003;5(5): 807-816.

38. Bland E. Nano ink Tatoo could monitor diabetes, 2009;[On-line]. Available: http://dsc.discovery.com/news/2009/02/12/diabetes-tattoo.html

39. Van Haeringen NJ. Clinical biochemistry of tears. Survey ofOphthalmology 1981; 26(2): 84-96.

40. Badugu R, Lakowicz J R, Geddes CD. A glucose sensingcontact lens: A Non-invasive technique for continuous physi-

ological glucose monitoring. J ournal of Fluorescence 2003;13(5): 371-374.

41. Badugu R, Lakowicz J R, Geddes CD. A glucose-sensingcontact lens: from bench top to patient. Current Opinion inBiotechnology 2005; 16: 100-107.

42. Gourzi M, Rouane A, Guelaz R, Alavi MS, McHugh MB, NadiM. Non-invasive glycaemia blood measurements by elec-tromagnetic sensor: study in static and dynamic bloodcirculation. J Med Eng Technol 2005; 29: 22-26.

43. Gourzi M, Rouane A, McHugh MB, Guelaz R, Nadi M.New biosensor for non invasive glucose concentrationmeasurement. Sensors, Proceedings of IEEE 2003; 2:1343-1347.

44. Cameron BD, Cot GL. Polarimetric glucose sensing in aque-ous humor utilizing digital closed-loop control. Engineeringin Medicine and Biology Society. Bridging Disciplines for

Biomedicine. Proceedings of the 18th Annual InternationalConference of the IEEE, 1996; 1: 204-205.

45. Boeckle S, Rovati L, Ansari RR. Polarimetric glucose sensingusing the Brewster-reflection off the eye lens: theoreticalanalysis. Proceedings of SPIE 2002; 4624: 160-164.

46. Rawer R, Stork W, Kreiner CF. Non-invasive polarimetric mea-surement of glucose concentration in the anterior chamberof the eye. Graefes Archive for Clinical and ExperimentalOphthalmology 2004; 242(12): 1017-1023.

47. Cot GL, Fox MD, Northrop RB. Noninvasive optical polari-metric glucose sensing using a true phase measurementtechnique. IEEE Transactions on Biomedica l Engineering1992; 39(7): 752-756.

non-invasive glucose sensors

Documents