Polarization sensitive optical low-coherence reflectometry for blood glucose monitoring in human subjects Polarization sensitive optical low-coherence reflectometry for blood glucose

Polarization sensitive optical low-coherence reflectometry for blood glucosemonitoring in human subjectsJitendra Solanki, Om Prakash Choudhary, P. Sen, and J. T. Andrews Citation: Rev. Sci. Instrum. 84, 073114 (2013); doi: 10.1063/1.4816018 View online: http://dx.doi.org/10.1063/1.4816018 View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v84/i7 Published by the AIP Publishing LLC. Additional information on Rev. Sci. Instrum.Journal Homepage: http://rsi.aip.org Journal Information: http://rsi.aip.org/about/about_the_journal Top downloads: http://rsi.aip.org/features/most_downloaded Information for Authors: http://rsi.aip.org/authors

REVIEW OF SCIENTIFIC INSTRUMENTS 84, 073114 (2013)

Polarization sensitive optical low-coherence reflectometry for bloodglucose monitoring in human subjects

Jitendra Solanki,1 Om Prakash Choudhary,1 P. Sen,2 and J. T. Andrews1,a)

1Applied Photonics Laboratory, Department of Applied Physics, Shri G S Institute of Technology and Science,Indore 452 003, India2Laser Bhawan, School of Physics, Devi Ahilya University, Khandwa Road, Indore 452 007, India

(Received 21 February 2012; accepted 5 July 2013; published online 26 July 2013)

A device based on polarization sensitive optical low-coherence reflectometry is developed to moni-tor blood glucose levels in human subjects. The device was initially tested with tissue phantom. Themeasurements with human subjects for various glucose concentration levels are found to be linearlydependent on the ellipticity obtainable from the home-made phase-sensitive optical low-coherencereflectometry device. The linearity obtained between glucose concentration and ellipticity are ex-plained with theoretical calculations using Mie theory. A comparison of results with standard clinicalmethods establishes the utility of the present device for non-invasive glucose monitoring. © 2013 AIPPublishing LLC. [http://dx.doi.org/10.1063/1.4816018]

I. INTRODUCTION

The optically active nature of glucose inspires us to usepolarization state of light as key for blood glucose monitor-ing. In an optically active medium, the velocities of right cir-cularly polarized (RCP) and left circularly polarized (LCP)lights differ. Consequently, such medium has the ability to ro-tate the plane of polarization of the resultant linearly polar-ized (LP) light.1, 2 Larger glucose concentration is expected toyield larger polarization rotation. Despite the wealth of differ-ent properties that can be probed with polarized light, the tur-bid nature of biological samples makes them non-trivial. Forexample multiple light scattering in turbid media can hindermeasurement and data interpretation. Over the years, many at-tempts were made to overcome these complexities involved inthe measurement of polarization properties of scattered lightfrom biological tissues.

Several techniques are proposed and adopted to inter-pret the data obtained from turbid media.3–13 Accordingto Ghosh et al.,3 the study of polarization properties ofbiological tissue using light scattering carries a wealth ofmorphological and functional information and has potentialbiomedical importance. They discussed comprehensive turbidpolarimetry platform consisting of forward Monte Carlo mod-eling and inverse polarization decomposition analysis. Heeet al.4 used optical coherence domain reflectometer for char-acterizing phase retardation between orthogonal linear polar-ization modes at each reflection point in a birefringent sam-ple. They used the reflectometer to characterize attenuationand birefringence of tunic media in calf coronary artery. Kohlet al.5 showed that presence of glucose dissolved in an aque-ous solution increases refractive index of the solution and alsoinfluences scattering properties of any particles suspendedwithin it. They demonstrated the effect of the glucose on lighttransport in highly scattering tissue simulating phantoms us-ing diffusion theory. Their results pave way to use this effect

a)Author to whom correspondence should be addressed. Electronic mail:[email protected]

for non-invasive glucose monitoring in diabetic patients. Thenext step in this direction was to correlate scattering coeffi-cient with blood glucose concentration. In 1997, Bruulsemaet al.6 examined reduced scattering coefficient of tissue in re-sponse to steep changes in blood glucose levels of diabeticvolunteers. They observed correlation between steep changesin blood glucose concentration and tissue reduced scatteringcoefficients in 30 out of 41 subjects. Wang et al.7 reportedexistence of linear relationship between angles of rotation inMuller matrix elements and concentration of glucose.

In a pilot study of specificity of non-invasive blood glu-cose monitoring using Optical Coherence Tomography (OCT)technique, Larin et al.8 report that several osmolytes maychange refractive index mismatch between interstitial fluid(ISF) and scattering centers in tissue. They found that the re-fractive index mismatch between the ISF and scattering cen-ters in tissue changes with several body osmolytes. However,the changes in refractive index mismatch due to glucose is oneor two orders higher than that obtained for other osmolytes.They concluded that OCT technique may provide blood glu-cose monitoring with sufficient specificity under normal phys-iological conditions. They also used phase-sensitive opticallow-coherence reflectometry (PS-OLCR) for measurement ofanalyte concentration9 and demonstrated a high degree of sen-sitivity and accuracy of phase measurements of analyte con-centrations in their experiment. Cameron and Li10 examinedthe polarization sensitivity of glucose concentration in phys-iological range for highly scattering biological media andconcluded that OCT based glucose monitoring is approach-ing standard invasive and minimally invasive techniques. Leeet al.11 measured changes of degree of circular polarization(DOCP) in intralipid suspensions and human cervical tissueusing polarization sensitive OCT (PS-OCT) system and pro-posed that this technique may provide a unique diagnostictool.

Motivated by the studies reported above, we have usedPS-OLCR technique for analyzing blood glucose concentra-tion in tissue phantom and in human subjects. We report the

0034-6748/2013/84(7)/073114/7/$30.00 © 2013 AIP Publishing LLC84, 073114-1

073114-2 Solanki et al. Rev. Sci. Instrum. 84, 073114 (2013)

measured values of ellipticity as a function of glucose con-centration in tissue phantom as well as in human subjects.The experimental observations have been explained usingMie scattering theory by incorporating the refractive indexdependence on the state of polarization in the optically activemedium.

II. EXPERIMENT

The optical low-coherence reflectometry setup developedby us is custom designed for the measurements with tissuephantom, blood samples as well as for non-invasive measure-ments with human subjects. A schematic of the Michelsoninterferometer type PS-OLCR is shown in Fig. 1(a). The un-folded view of the optical components and the state of po-larization of light after passing through each of them in sam-ple and reference arms are shown separately in Figs. 1(b) and1(c), respectively. A low coherence broadband light source,viz; Hamamatsu Superluminescent diode (SLD) having a cen-ter wavelength of 835 nm and bandwidth 50 nm is used.The sample arm of the device has the option to accommo-date samples of tissue phantom, blood or any one finger ofthe subject for non-invasive measurements. As exhibited inFigs. 1(a) and 1(b), low coherence beam of light emitted fromSLD was passed through the linear polarizer and a quarterwave plate at 45◦ to the polarizer axis as shown in the fig-ure to provide LCP light. The LCP beam was delivered to a50-50 non-polarizing beam splitter (BS). The reference armwas scanned repetitively at a constant speed of 3 mm/s. FromFigs. 1(a) and 1(c) one can notice that the light behaves iden-tically in both sample and reference arm till the beam splitter.After that, the light is guided to the sample via a mirror at45◦ to the incident beam. The sample is assumed to changethe polarization state of the medium. Accordingly, for theo-retical understanding, we assume the sample to behave likea combination of optical rotator (WS) and a mirror. The sig-nal backscattered from the sample will have mixed state ofpolarization. The back reflected light from the reference armand the backscattered light from the sample arm form an in-terferogram, which is delivered to the polarizing beam splitter(PBS). The orthogonal components of the interference signalare detected by two photo diodes D⊥ and D‖. These orthogo-nal components of the OLCR signals were filtered, amplified,and delivered to a laptop for further processing and analysis.

A. Data collection

The sample arm of the device consists of a reference glassplate. As shown in Fig. 1, the sample under study makes directcontact with the top surface of the glass plate. The OLCR sig-nal obtained from samples such as tissue phantom or humansubject is expected to give two distinct peaks, correspondingto the air/glass and glass/tissue interfaces. The signal obtainedfor air/glass interface is a Gaussian having signatures of thelight source while the signal obtained for glass/tissue inter-face have the signatures of the sample under study. Since theoptically active medium rotates the LCP and RCP at differ-ent magnitudes, finite ellipticity occurs for different values of

(a)

(b)

(c)

M

FIG. 1. (a) Schematic of the polarization sensitive optical low-coherence re-flectometry setup: SLD – superluminescent diode, P – polarizer, W – quar-ter wave plate, BS – non-polarizing beam splitter, S – sample, R – refer-ence mirror with scanning assembly, PBS – polarizing beam splitter, D, D⊥,D‖ – photo diodes, DAQ – data acquisition system, ED – envelope detector,PC – laptop, and other electronic processing units. The lower corner showsthe photograph of the compact experimental setup developed and used by theauthors. (b) The schematic shows the unfolded view of light propagation inthe reference arm. The state of polarization of light after each optical compo-nents is also shown. (c) The unfolded view of light propagation in the samplearm is shown here. The sample is assumed to behave as an optical rotator,accordingly, it is shown as a combination of mirror and a waveplate. Hence,the backscattered light from the sample may have mixed state of polarizationdue to additive nature, while the reflected light from the sample retains thestate of polarization.

glucose concentrations. Accordingly, the PS-OLCR setup em-ploys a PBS to disentangle the orthogonal components of thelight polarization. The amplitudes of the backscattered sig-nal from the sample are measured using the photo diodes D⊥and D‖.

B. Sample preparation: Tissue phantom

Intralipid is widely used in many optical experiments tofind the scattering properties of biological tissues and can beused as a good scatterer like RBCs in human blood. Intralipidis an emulsion of soy-bean oil, egg phospholipids, and glyc-erin. The major advantages of intralipid are its well knownoptical properties and the similarity of its microparticles tolipid cell membranes and organelles that constitute the sourceof scattering in biological tissue. We used intralipid as a tis-sue phantom that provides the backscattered component of an

073114-3 Solanki et al. Rev. Sci. Instrum. 84, 073114 (2013)

incident light. Average size of scatterers in intralipid mea-sured using laboratory confocal microscope is found to be3.5 μm with refractive index of 1.42.

The measurement with tissue phantom was taken un-der the hypoglycemic, normal, and hypoglycemic glucoselevels of humans as (<80 mg/dl), (80–110 mg/dl), and(>110 mg/dl), respectively. Following measurement modalityis adopted during the experiment for use with tissue phantom.First, a fixed volume (1 ml) of intralipid (20% w/V) was di-luted to 100 ml using distilled water and diluted solution of0.01% of tissue phantom was prepared. In the second step,200 mg of glucose was dissolved in 100 ml distilled waterto make glucose solution. In the third step 100 μl of glucosesolution was added to 1 ml of tissue phantom using eppen-dorf micropipette. This gives us 20 mg/dl glucose concentra-tion. Two minutes of settlement time was given between tissuephantom and every glucose concentrations. The glucose con-centration was increased upto 200 mg/dl. Only 10 μl of wholesolution was placed on 140 μm thick cover-slip in the samplearm and five sets of measurements of backscattered light weretaken for each concentration.

The results obtained from the intralipid were verified us-ing Mie scattering theory. We have further extended the anal-ysis with voluntary human subjects.

C. Sample preparation: Non-invasive measurements

Our PS-OLCR device has the flexibility of monitoringblood glucose invasively as well as non-invasively. In thepresent case, the thumb finger of the subject is used in thesample arm instead of tissue phantom in the previous case.Before placing the finger, the distal pulp region of thumb wascleaned by mild spirit. A thin layer of glycerin was appliedfor better optical contact between thumb skin of human sub-ject and cover-slip. Measurements were performed simultane-ously using PS-OLCR setup and with standard glucometer orthrough chemical route (GOD/BOD Method).

It is well known that after a meal, a temporary rise inblood sugar occurs. The extent and duration of sugar level de-pends on the characteristics of the food as well as the subjects.For a normal subject, the increase in blood sugar level may notexceed 160-180 mg/dl. In case of diabetic subject, the rise andreduction of glucose level after meal is slower and elevated.The subjects volunteered in the current experiment are askedto consume 100 g glucose with water. After a settling time of10 min, the measurement of blood glucose were carried outusing PS-OLCR system and through glucometer, simultane-ously. The measurement were repeated for every 10-15 mininterval. A correlation between the glucose concentration andellipticity of the polarized light obtained and analyzed.

III. THEORETICAL ANALYSIS

A theoretical model is developed to understand the exper-imental observations. Intralipid and human blood can be the-oretically categorized as turbid media in which the backscat-tered light from sample arm contributes to OLCR signal. Thesignal from the sample arm comprises of single and multiple

scattered components of light. Strong multiple scattering typ-ical of most biological tissues leads to loss of phase, directionand polarization of incident radiation. It has been observedthat for circularly polarized light randomization of polariza-tion requires more scattering events than the randomizationof its direction.14–16 Multiply scattered polarized light froma turbid medium carries information about refractive indexvariation. The incident light in the present experimental setupis left circularly polarized. Tinoco and Keller17 discussed dif-ferential scattering of circularly polarized (CP) light and con-cluded that circular differential scattering retains informationconcerning the chiral properties of the scattering object evenwhen the sample is partially or completely disordered. In anoptically active medium, the circularly polarized electromag-netic wave propagates without the change in the state of polar-ization. We consider, single scattering events of the CP lightby the scatterers in a chiral medium. Integration over all scat-tering angles yields the average polarization characteristics ofthe scattered light. The expressions for the incident Ei andscattered field Esca are given by18

Ei =∞∑

n=1

En

(M(1)

oln − iN(1)eln

)(1)

and

Esca =∞∑

n=1

En

[ianN(3)

eln − bnM(3)oln + cnM(3)

eln − idnN(3)oln

],

(2)respectively, with En = inE0

2n+1n(n+1) , E0 is the amplitude of the

incident electromagnetic wave, an, bn, cn, and dn are the coef-ficients of the scattered field while Nijk and Mijk are the spher-ical harmonics corresponding to the scattered field.18 Thecoefficients of the scattered field are functions of the relativerefractive index of the scatterer and the fluid. In the presentcase, RBCs are the scatterers while the extra cellular fluid(ECF) is the fluid. In general, the change in the concentra-tion of solute (c) changes the relative refractive index μ givenby

μ(c) = μ0 + dμ

dcc, (3a)

where μ is the refractive index of the solvent.19 The normalmodes of the glucose solution are right (+) and left (−) circu-larly polarized modes. Consequently, in the present situationthe refractive indies for these mode can be written as

μ±(c) = μ0 ± dμ

dcc. (3b)

In the optically active medium, the polarization rotation takesplace and the corresponding rotatory power (ρ) is given by20

ρ = π

λ|μ+(c) − μ−(c)| = 2π

λ

dμ

dcc. (4)

The rotatory power represents the rotation of the polarizationplane per unit length. The above equation confirms that themagnitude of change in the state of polarization is a functionof the distance travelled by the light in the active medium asshown in Figure 1 and also on the concentration of sugar inthe medium. The concentration dependent change in refrac-tive index dμ/dc are measured and discussed in Sec. IV while

073114-4 Solanki et al. Rev. Sci. Instrum. 84, 073114 (2013)

the distance travelled by light are accounted by single andmultiple scattering components within the coherence lengthof light source. Hence, the amplitude of backscattered com-ponents of light will be a mixture of various phase delayedleft circular polarized components represented as

Esca =∞∑

n=1

∑j

En

[ianN(3)

eln − bnM(3)oln + cnM(3)

eln

− idnN(3)oln

]× exp

[ik

(μ0 − dμ

dcc

)Lj

], (5)

where, k is the wave-vector, Lj is the path length travelled bylight for different scattering events and the ellipticity will bedecided by additive nature.

The coherent gating signal detection method adopted inthe present setup removes the DC components of the interfer-ence signal and the signal envelope detected are representedby21

Id (τ ) = 2Re{E∗

s (t) · Er (t + τ )}R(τs)

1/2 ⊗ C(τs), (6)

where s and r, represent the sample and reference arms, re-spectively. Ls(r) is the distance of sample (reference) arm fromthe beam splitter cube. τ = [τ s − τ r = (Ls/c − Lr/c) = �L/c]is the time delay corresponding to the round-trip optical pathlength between two beams and Es(t) is the path length re-solved field amplitude while Er is field amplitude from thereference mirror.

R(τ s) and C(τ s) are the path length resolved diffuse re-flectance and low coherence function of the source, respec-tively. After coherent gating the maximum amplitude of mea-sured OLCR signal at the detector would be 2Re(E∗

s Er ). Toaccount for the single scattering events occurring at the inte-rior depth of the tissues as well as to improve the signal qual-ity the signal Id(τ ) is integrated over a length up to subcutiscells of tissue.

As shown in Fig. 1, we launch a left circularly polarizedlight into the sample and reference arms. The horizontal andvertical components of the interfering signals are detected us-ing polarizing beam splitter cube and two detectors D‖ andD⊥. Accordingly the signal amplitude as measured by the de-tectors after signal processing is given by

IDp =[∫ τ2

τ1

Id (τ )dτr

]p

, (7)

p = ⊥ and ‖, τ 1 and τ 2 are the limit of numerical integra-tion corresponding to the positions L1 and L2 of the scatteredsignal from reference plate upto the tissue layer as discussedin Sec. III A. The backscattered component from the samplearm comprises of the snake, ballistic, and diffuse componentscorresponding scattering sites. Normally in an OLCR exper-iment these signals lying within the coherence lengths aredetected. Accordingly, we consider the backscattered compo-nent as a mixture of linear and circular polarized components.The mixture of the differently polarized state will be splittedby the PBS to ID⊥ and ID‖ components. The Stokes param-eters (Q, U, V ) for these components are additive and the

FIG. 2. Estimated values of ellipticity against the angle of optical rotation inthe sample arm. These estimations were made using Jones matrix calculationusing the schematic of Figs. 1(b) and 1(c). The inset showing linear changeis a magnified view.

ellipticity (η) is given by18, 22

η = V√Q2 + U 2

= ID⊥ID‖

. (8)

The above expression, in conjunction with Eqs. (1)–(7),shows that the ellipticity will have signatures of glucose con-centration in the sample under study.

In order to verify the additive nature of polarization stateof the sample, we have performed theoretical estimation ofellipticity (η) from

η = (Es)⊥(Er )⊥(Es)‖(Er )‖

, (9)

where Es = JBSJMJWSJSJWS

JMJBSJWJP E0 and Er = JBS

JRJBSJWJP E0. Ji is the Jones matrix [Ref. 20, pp. 197-230]the ith component shown in Figures 1(b) and 1(c). The esti-mated variations in ellipticity with the change in the orienta-tion of the quarter wave plate is shown in Figure 2. The curveindicates a linear increase for smaller variations in optical ro-tation, while for large angle of optical rotation it exhibits asinusoidal nature. In case of biological samples, we expectsmall angle of optical rotation therefore, in the inset the mag-nified view of same curve for smaller variation in optical ro-tation is given.

A. Signal processing

We have designed and developed our own PS-OLCRsetup, hence it has lots of flexibility in data acquisition. OLCRsignals are obtained from both detector D⊥ and D‖, simul-taneously. The typical OLCR signal obtained from the de-tector D⊥ is shown in Fig. 3(a). The signal exhibits a sin-glet from the lower side of the cover slip and a doublet fromthe glass-finger interface. The sample holder (S) shown inFig. 1, is tilted slightly such that the signal for glass-tissueinterface is larger while the signal from air-glass interfaceremains smaller. The boxed regions in Fig. 3(a) are magni-fied and the two peaks are shown in Fig. 3(b). The separationbetween the peaks are �223 μm, which is attributed to the

073114-5 Solanki et al. Rev. Sci. Instrum. 84, 073114 (2013)

FIG. 3. OLCR signals obtained from the reference plate (cover-slip of thick-ness μm) and sample are shown. The complete signal is shown in (a) at top,where a-g and g-t represent the air-glass and glass-tissue interface, respec-tively. The planes from the signal obtained are shown schematically. The sig-nal obtained from the tissue-glass interface is magnified and shown in (b).The signal is integrated (shaded region) up to a distance of 400 μm ignoringthe coherent signal obtained for the glass surface is shown in (c) at bottomcorner. The integrated signal is used while estimating ellipticity of the polar-ized light.

thickness of the cover-slip of 150 μm and the refractive indexof 1.55. Figure 3(c) shows the signal obtained at the glass-tissue interface. The signal is integrated numerically upto amaximum of 50 data points (from positions L1 and L2 asshown in Fig. 3(c)), corresponding to a depth of ≈400 μm.From available literature23 this depth corresponds to the epi-dermis, dermis and parts of subcutis tissues.

The numerical integration D⊥ to a depth of 400 μm iskept constant for all subjects. Similar signal processing is car-ried out for the signal obtained from D‖. The time delay be-tween the two measurements ID⊥ and ID‖ is less than 100 ms.Using Eq. (8), the ellipticity is estimated and saved for furtherprocessing. The measurements were repeated many times toconfirm reproducibility and to remove any spurious data.

The experimental setup was tested for its polarizationsensitivity. For this purpose, the sample is replaced with a mir-ror and a quarter wave plate. Initially the QWP is aligned to itsfast axis and angle of rotation is considered zero. The QWP isrotated from its fast axis to slow axis and further. Correspond-ing ellipticity data were taken for various angles of the QWP.The observations are shown in Figure 4. The solid squares inthe figures are the experimental data points while the solidline represents the theoretically calculated data. One can eas-ily recognize the excellent agreement between theoretical cal-culations and the experimental observations from the figureensuring the validity of polarization sensitive measurementsof the setup.

In human subjects the major contribution of light scatter-ing comes from the red blood cells. Accordingly, the averagesize of the scatterers (RBC) are considered as a = 7 μm.

The change in the refractive index of ECF can also oc-cur due to the variations in the concentrations of variouselectrolytes present in the human blood. In order to analyzeindividual effect of each of them on the refractive index varia-

FIG. 4. Measured values of ellipticity with angular position of waveplate.The results matches with the theoretical values as shown in Figure 2.

tion, we have used standard Abbe refractometer. As describedearlier in the Sec. II B, the tissue phantom is prepared withvarious glucose as well as various electrolyte concentrationsand the measured values of refractive indices are exhibitedin Fig. 5. The curves a, b, c, and d demonstrate the depen-dence of refractive index with the concentration of glucoseand other electrolytes KCl, Urea, NaCl, respectively. Shadedrectangles on the curves indicates the normal values of respec-tive electrolyte.23 The variation in the refractive index withinthe dashed horizontal lines corresponds to the refractive in-dex for normal range of glucose, and electrolytes concentra-tion for healthy human subject.23 A value of refractive indexabove or below this limit shows the hyper or the hypo regions,respectively. This figure further reveals that the variation in re-fractive index with change in glucose concentration dμ/dc isvery large (curve a) as compared to that measured for otherelectrolytes (curves b-d). The change in glucose concentra-tion will also affect the rotatory power and will reflect in thechange in the ellipticity of the signal.

FIG. 5. Change in refractive index with concentration of glucose (a) andother electrolytes (b-d). The boxed regions show the normal body values ofhuman subjects.23

073114-6 Solanki et al. Rev. Sci. Instrum. 84, 073114 (2013)

FIG. 6. Theoretical (solid line) and experimental (filled dots) observationsof ellipticity with increasing glucose concentration using tissue phantom isexhibited. The fluctuations in experimental data is attributed to the size dis-tribution of scatterers in the tissue phantom.

IV. RESULTS

The analytical understanding of ellipticity obtainablefrom Eq. (8) is not trivial. Hence, to understand the responseof glucose concentration with ellipticity, we have made nu-merical estimation of ellipticity using Eqs. (1)–(8) and exhib-ited in Fig. 6 (as solid line). The results show nearly no changeof η at low concentration of glucose, while notable changesare observed at large values of glucose concentration. To sup-port the theoretical and numerical estimations, we performedan experiment for the measurement of η with tissue phantomas light scatterers. The measured values of ellipticity for dif-ferent values of glucose concentration are plotted in Fig. 6 (asfilled circles). The experimental and theoretical data followsimilar trend that for no change in ellipticity to large changein ellipticity is observed with change in glucose concentra-tion.

Armed with the encouraging results obtained for the sam-ples of tissue phantom, we proceed further with human sub-jects. The ellipticity obtainable from ID⊥/ID‖ is plotted inFig. 7 as a function of glucose concentration. Figs. 7(a) and7(b) exhibit two sets of measurements each obtained fromsubjects 1 and 2, respectively. In order to verify the repro-ducibility of measurement, the data for each subject weretaken on different days shown by filled and open circles inthe figure. The solid curve is a linear fit to the experimentaldata.

The automated program is designed to collect at least5 data points each from detectors D⊥ and D‖. Error bars wereobtained after finding various values of η and error analy-sis. When the measurements are repeated on different days,similar procedure is followed, and the values found to fol-low the linear curve with same slope. It is interesting to notehere that the slope (ratio of ellipticity and glucose concen-tration) remains a constant for a particular subject and con-firms the repeatability of the measurements. Similar resultswere obtained for other subjects and shown in Fig. 7(c). Thecalculated slope obtained for different subjects are given inTable I.

FIG. 7. The value of ellipticity measured using the device developed by theauthors group (Y axis) and the simultaneous measurement of glucose con-centration using commercial grade glucometer (X axis) are shown. The dataobtained from subjects 1 and 2 are shown in (a) and (b). The filled circle cor-responds to first set of measurements while the open circles, second set ofmeasurements. (c) The ellipticity values obtained with six subjects. The linesare drawn to fit the experimental data.

TABLE I. Statistics and slope values of the subjects.

Slope Age RBC SizeSubject (mg/dl)−1 (Years) (μm)

Subject - 1 0.01844 27 8.12Subject - 2 0.02138 25 7.52Subject - 3 0.03158 26 8.12Subject - 4 0.03965 32 8.10Subject - 5 0.05442 30 7.90Subject - 6 0.13437 29 8.20

073114-7 Solanki et al. Rev. Sci. Instrum. 84, 073114 (2013)

FIG. 8. Measured values of ellipticity of polarized light (with error bar) andglucose concentration (solid squares) with time after a glucose drink is ex-hibited here. Both the curves follow closely.

In order to establish the validity and the utility of the mea-surements reported here, we have also taken the time variationof glucose concentration in blood and have compared it withthe standard chemical method (GOD/BOD). For this purpose,we examined the subject for variation in blood glucose levelin every 10-15 min interval on both, the PS-OLCR setup andglucometer after a glucose drink.

In normal subjects the blood glucose level stabilizeswithin 120 min. It is well known24 that the blood sugar levelin human subjects initially increases after taking meal for30–45 min. In some of the subjects, a hypoglycemic dip mayalso occur. In order to demonstrate the PS-OLCR as a non-invasive glucose monitor, the glucose level variation as afunction of time has been plotted in Fig. 8. The glucome-ter measurements are shown as a solid square while elliptic-ity measurements are shown by open squares with error bar.We find an excellent quantitative agreement between thesecurves.

V. CONCLUSIONS

We have designed and developed PS-OLCR for non-invasive glucose monitoring of blood glucose with humansubjects. The results exhibited in Fig. 7 shows that a definitelinear correlation exists between the glucose concentrationand the ellipticity. We could find that the ratio of ellipticityand glucose concentration is a reproducible constant value fora person. It is further established that the excellent agreementreported in Fig. 8 between ellipticity and glucose concentra-tion as measured using PS-OLCR and GOD/BOD method canbe a future tool for noninvasive glucose monitoring device.

ACKNOWLEDGMENTS

The authors thankfully acknowledge the support andvaluable time offered by the volunteers. The authors thankProfessor P. K. Sen, SGSITS for fruitful discussions. Theyalso thank the learned referee for the suggestions, which hasimproved the quality content of the paper. They also acknowl-edge the financial support received from UGC and ICMRNew Delhi, India and MPCOST, Bhopal, India.

1M. Born and E. Wolf, Principles of Optics: Electromagnetic Theory ofPropagation, Interference and Diffraction of Light, 7th ed. (CambridgeUniversity Press, Cambridge, 1999), pp. 647–661.

2A. Ishimaru, Wave Propagation and Scattering in Random Media (Aca-demic Press, New York, 1978).

3N. Ghosh, J. Soni, M. F. G. Wood, M. A. Wallenberg, and I. A. Vitkin,Pramana, J. Phys. 75, 1071–1086 (2010).

4M. R. Hee, D. Huang, E. A. Swanson, and J. G. Fujimoto, J. Opt. Soc. Am.B 9, 903–908 (1992).

5M. Kohl, M. Cope, M. Essenpreis, and D. Bocker, Opt. Lett. 19, 2170–2172(1994).

6J. T. Bruulsema, J. E. Hayward, T. J. Farrell, M. S. Patterson, L.Heinemann, M. Berger, T. Koschinsky, J. S. Christiansen, H. Orskov,M. Essenpreis, G. S. Redeker, and D. Bocker, Opt. Lett. 22, 190–192(1997).

7X. Wang, G. Yao, and L. V. Wang, Appl. Opt. 41, 792–801 (2002).8K. V. Larin, M. Motamedi, T. V. Ashitkov, and R. O. Esenaliev, Phys. Med.Biol. 48, 1371–1390 (2003).

9K. V. Larin, T. Akkin, R. O. Esenaliev, M. Motamodi, and T. E. Milner,Appl. Opt. 43, 3408–3414 (2004).

10B. D. Cameron and Y. Li, J. Diabetes. Sci. Technol. 1, 873–878 (2007).11S. W. Lee, J. H. Kang, J. Y. Yoo, and B. M. Kim, in Proceedings of the

29th Annual International Conference of IEEE Engineering in Medicineand Biology Society, France (IEEE, 2007), pp. 3350–3352.

12R. Poddar, S. R. Sharma, J. T. Andrews, and P. Sen, Curr. Sci. 95, 340–344(2008).

13J. Solanki, P. Sen, J. T. Andrews, and K. K. Thareja, J. Mod. Phys. 3, 64–68(2012).

14W. Cai, X. Ni, S. K. Gayen, and R. R. Alfano, Phys. Rev. E 74, 056605-10(2006).

15X. Ni and R. R. Alfano, Opt. Lett. 29, 2773–2775 (2004).16M. Xu and R. R. Alfano, Phys. Rev. Lett. 95, 213901-1–213901-4 (2005);

Phys. Rev. E 72, 065601(R) (2005).17I. Tinoco, Jr. and D. Keller, J. Phys. Chem. 87, 2915–2917 (1983).18C. F. Bohren and D. R. Huffman, Absorption and Scattering of Light by

Small Particles (John Wiley, New York, 1998).19J. S. Maier, S. A. Walker, S. Fantini, M. A. Franceschini, and E. Gratton,

Opt. Lett. 19, 2062–2064 (1994).20B. E. A. Saleh and M. C. Teich, Fundamentals of Photonics, 2nd ed. (John

Wiley, New Jersy, 2007), p. 228.21C. Akcay, P. Parrein, and J. P. Rolland, Appl. Opt. 41, 5256–5262

(2002).22M. Bass, E. W. V. Stryland, D. R. Williams, and W. L. Wolfe, Handbook of

Optics (McGraw-Hill, 1995), Vol. II, Chap. 22.23H. Marcovitch, Black’s Medical Disctionary, 41st ed. (A & C Black, Lon-

don, 2005), pp. 804–805.24H. Varley, Practical Clinical Biochemistry: Glucose Tolerance Tests and

Tests for Investigating Hypoglycemics, 4th ed. (Manchester University,1969), pp. 97–109.