Sb

JL

a

ARRAA

KL(UI

1

asiic

c1ecppc

cncdssiat

0d

Talanta 79 (2009) 352–360

Contents lists available at ScienceDirect

Talanta

journa l homepage: www.e lsev ier .com/ locate / ta lanta

trategies for the identification of urinary calculus by laser inducedreakdown spectroscopy

esús Anzano ∗, Roberto-Jesús Lasherasaser Laboratory and Environment, Department of Analytical Chemistry, Faculty of Sciences, University of Zaragoza, Pedro Cerbuna #12, 50009 Zaragoza, Spain

r t i c l e i n f o

rticle history:eceived 15 January 2009eceived in revised form 23 March 2009

a b s t r a c t

The present work studies two different strategies to identify urinary calculus. On one hand, (linear orparametric and rank or non-parametric) correlation methods using a �-LIBS system are studied. On theother hand, elemental ratios of reference materials are determined by using a higher-energy laser and

ccepted 27 March 2009vailable online 5 April 2009

eywords:aser induced breakdown spectroscopyLIBS)

an Echelle spectrograph with an ICCD camera, although without microscope. A data-treatment methodwas applied for each system and real samples of kidney stones – previously analyzed by IR spectroscopy– were used for reliable evaluation of two identification strategies.

© 2009 Elsevier B.V. All rights reserved.

rinary calculusdentification

. Introduction

Stone formation in urinary tracts or kidney is known as urolithi-sis. About 4% of Spanish women and men develop a kidney stone atome time in their life—an increasing prevalence in both sexes. Itsncidence rate is 0.27% in Spain; therefore, there are 1.7 millions ofnhabitants suffering from kidney-stone disease and 100,000 newases appear every year [1].

Approximately 75% of these stones are usually composed of cal-ium sales (50% of these stones are calcium oxalate, CaOx) and0–20% are composed of uric acid (UA). In Zaragoza (Spain), anpidemiology study showed that 40% of stones are composed ofalcium oxalate (CaOx), 30.6% of calcium oxalate and calcium phos-hate (CaP) mix, 21.7% of uric acid, 0.9% of magnesium ammoniumhosphate (MgP), 0.9% of cystine and 5.9% of others (sodium urate,alcium carbonate, etc.)

Therapy to prevent calculi recurrence requires estimating theomposition of urinary calculi. Traditional wet chemistry tech-iques, X-ray diffraction and infrared (IR) spectroscopy are theurrent analytical methods [2]. These techniques include someisadvantages; for instance, wet chemical analysis requires largeample amounts, while X-ray diffraction cannot detect amorphous

amples. IR spectroscopy has been applied within clinical chem-stry but often provides complex spectra with contributions fromsizeable number of unknown interfering substances [2]. IR spec-

roscopy has often been combined with complex software such as

∗ Corresponding author. Tel.: +34 976762684; fax: +34 976761292.E-mail address: janzano@unizar.es (J. Anzano).

039-9140/$ – see front matter © 2009 Elsevier B.V. All rights reserved.oi:10.1016/j.talanta.2009.03.065

multivariate calibration methods. Thus, Fourier transform infraredspectroscopy (FT-IR) has been used to perform a computerizedsearch in several libraries and partial least-squares (PLS) regres-sion has been tested for the most frequent urinary compositions.However, these techniques require several steps of data prepro-cessing [3]. To minimize data treatment, artificial neural networkmodels for detecting the most frequent stone compositions havebeen studied [4] and a single reflection accessory for FT-IR and neu-ral network algorithm interpretation has been developed; however,visual inspection of spectra is still necessary [2].

Laser induced breakdown spectroscopy (LIBS) or Laser inducedplasma spectrometry (LIPS) has been used for clinical analysis [5–8].LIBS’ advantages for clinical analysis are: sample treatment is notnecessary and little amount of sample is used; this technique isnot destructive and, therefore, the sample may be re-analyzed bymeans of other techniques; and it is very fast and the sample canbe studied by using microscope when the �-LIBS system is used.

LIBS has been used to identify organic compounds by means ofdifferent strategies: (i) correlating the LIBS spectrum of unknowncompounds with library spectra [9,10], (ii) determining the ratios ofdifferent emission lines and molecular bands [11–16], and (iii) usingartificial neural networks [17]. The parametric and non-parametriccorrelation method involves studying multiple emission lines andmolecular bands. However, methods working with determiningratios almost exclusively consist of a few lines and bands.

Although limitations for identification exist – due to the loss ofmolecular information in plasma – the technique achieves excel-lent potential for online, real-time analysis. Nd:YAG laser has beenused for other applications due to its simplicity, easy operation,high efficiency, low cost and suitability. Previous papers [9,10] have

J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360 353

y ston

safiacaoHtoL[ks

ppOmw

m

Fig. 1. Kidne

hown that simple statistical correlation methods – such as linearnd rank correlations – can be successfully applied for the identi-cation of solid and particulate materials. Other papers prove thepplication of the LIBS technique for elemental analysis of renalalculus with concentrations of majority elements both in percent-ges (Ca and Mg) and ppm (Na, K, Sr, etc.) [19], and for analysisf the spatial distribution of these elements [20] in renal calculus.owever, results on the different elements are only at quantita-

ive level and these works do not clearly study the applicationf LIBS for renal-calculus identification. These papers reveal thatIBS offers a suitable method to obtain quantitative information19,20] on spatial distribution [20] of elements in the differentinds of stones found in the human body, without destroying thetones.

Two different strategies and instruments are studied in theresent work. On one hand, (linear or parametric and rank or non-arametric) correlation methods using a �-LIBS system are studied.

n the other hand, elemental ratios of reference materials are deter-ined by using a higher-energy laser and an Echelle spectrographith an ICCD camera, although without microscope.

The aim of the first strategy is identifying urinary calculus byeans of a �-LIBS by using a spectrum library, and linear and rank

es analyzed.

correlation. This method involves studying all emission lines andmolecular bands appearing in spectra from 200 to 850 nm. Theobjective in the second strategy is to identify kidney stones by deter-mining intensity ratios from elemental lines and measuring ratiosof several elements such as C, Ca, H, Mg, N, O and P.

2. Experimental

2.1. Reactives and samples

The library was prepared from the most frequent composition ofurinary calculi by using calcium oxalate, calcium phosphate, cystine(Scharlau, extra pure), uric acid (Panreac, PRS) and mixes (calciumphosphate and magnesium phosphate 1:1, calcium oxalate andcalcium phosphate 1:1) as reference materials. A pellet of each com-pound and mix was prepared. Standards (0.50–0.75 g) were mixedwith a mortar and pressed at 105 N for 5 min.

Real samples of kidney stones provided by the University Hospi-tal of Zaragoza were analyzed. Some samples were broken open toanalyze their interior part, since stone core may differ from stoneexterior; stone core is the reason for the occurrence of urinary cal-culus. Photographs of the most frequent stones are shown in Fig. 1.

354 J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360

-LIBS

As

2

2

UutcwOFsi2fi3o22

13.6 × 13.6-�m2/pixel, 18-mm intensifying diameter and exposure

Fig. 2. �

ll real samples of kidney stones were previously analyzed by IRpectroscopy.

.2. Instrumental setup

.2.1. Strategy A: correlation methodsThe equipment consisted of a Nd:YAG laser (Quantel, model

ltra CFR) coupled to a microscope; a 30× focalization lens wassed. Pulse energy was approximately 9 mJ. Both laser and spec-rometer are synchronized by a trigger pulse from a home-madeompact pulse generator. Radiation from the laser spark is collectedith a bifurcated optical fiber connected to a dual-channel Ocean-ptics mini spectrometer (SD2000, Ocean Optics Inc., Dunedin,L, USA). Plasma was collected in non-collinear mode. The �-LIBSystem used is shown in Fig. 2. The position of the collimat-ng lens (74-UV Ocean Optics, it has an f/2 fused silica lens for00–2000 nm, 5-mm diameter, 10-mm focal length) was adjustedor maximum light collection. The spectrometer has the follow-

ng characteristics: channel one (slave), 230–310 nm spectral range,600 mm−1 holographic grating, 25 mm slit, 0.3 nm spectral res-lution, 2048 pixel linear CCD array; and channel two (master),00–850 nm spectral range, 600 mm−1 grating blazed at 400 nm,5 mm slit, 1.3 nm spectral resolution, 2048 pixel linear CCD array.

Fig. 3. Methodological scheme u

system.

Spectrometer is driven from a laptop computer (hp invent, Omni-book XE3) via a DAQCard—700 interface (National Instruments,USA).

2.2.2. Strategy B: ratios methodThe remainder instrumentation used consisted of a Nd:YAG

laser, an xyz stage carrying the sample (Standa 011957), a spectro-graph and an intensified charged coupled device (ICCD) detector.A Nd:YAG laser (Brilliant Quantel, Q-Switched) with a 115 mJ laserpulse energy at the second harmonic of 532 nm, a 4.4 ns pulse dura-tion and 0.7 Hz repetition frequency was used with a 90-mm focallength lens. Plasma light was collected and transported to spectro-graph by lens and optical fiber (fused silica, 50-�m core diameter).An external collector/collimator was used as collected lens; its posi-tion was adjusted at 200 mm by a diode laser (Andor, HE-OPI-0009).An Echelle spectrograph (Andor Mechelle ME5000, 195-mm focallength, F/7, �/�� 5000, spectral range from 200 to 975 nm) was cou-pled to an ICCD detector (Andor iStar DH734, 1024 × 1024-pixels,

sing correlation methods.

time was 0.011 s). This system was calibrated by a Hg:Ar lamp(Ocean Optics, HG-1, Hg–Ar lines 253–922 nm). MCP gain was fixedat 180. A 1-�s measurement gate delay time and a 1-�s integrationtime were used.

J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360 355

(A) C

2

mmt

mciditcli(4bn

Fig. 4. LIBS spectra of reference materials used for the library:

.2.3. Methodology with strategy AOptimization was carried out by means of collector-lens adjust-

ent by using a diode laser and microscope diaphragm widthaximizing the signal value. Delay time could not be modified due

o the home trigger used; therefore, delay time was fixed.The spectra library was prepared with pure compounds and

ixes (CaOx, CaP, MgP, UA, CY and CaP MgP mix). Five repli-ates were analyzed for each compound and spectra average wasntroduced as reference spectrum. Samples were analyzed at fiveifferent points. A methodology scheme for this strategy is shown

n Fig. 3. If a metal (Ca, Mg or Na) is not present in the LIBS spec-ra, the sample may be UA or CY and is identified by its correlationoefficient. If a metal (Ca, Mg or Na) is present and its emissionines predominate in the spectrum (weak emission lines are not

ndicative), there are three possibilities, if this metal is (a) calciumits emission lines are: 317.82, 393.26, 370.83, 373.6, 393.26, 396.69,30.18, 443.41, 445.45, 526.52 and 559.91 nm), then the sample maye CaOx or CaP and correlation methods must be applied; (b) mag-esium (its emission lines are: 279.69, 285.36, 293.50, 382.93 and

Y, (B) UA, (C) CaP, (D) MgP, (E) CaOx and (F) CaP and MgP mix.

518.30 nm), the sample may be MgP; and (c) sodium (its emissionlines are: 589 and 820.36 nm), the sample may be sodium urate. Lin-ear (parametric) and non-parametric (rank) correlation methodswere applied in this LIBS system for correct compound identifi-cation, independently from the previously explained methodology,which is only used as a first indicator. Besides, two spectral windows(200–400 and 200–850 nm) with the same resolution were studieddue to interferences (atmospheric nitrogen and oxygen) appearingat high wavelengths.

Simple application software was designed for data treatment. Itcalculates average spectrum, stores the spectra library and calcu-lates both linear and rank correlation coefficients.

2.2.4. Methodology with strategy B

Delay time was optimized. In this optimization carbon

(247.9 nm) and hydrogen (656.2 nm) lines were studied dependingon signal/noise value.

The second strategy consists on calculating C:Ca:H:Mg:N:O:Pratios in reference samples and thus the critical values of these

356 J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360

Table 1Emission lines and bands using �-LIBS.

Wavelength (nm)

Calcium 316.06 (w,II); 317.82 (m,II); 370.83 (w,II); 373.6 (w,II); 393.26 (s,II); 396.69 (s,II); 430.18 (w,I); 443.41 (w,I); 445.45 (w,II); 526.52 (w,I); 559.91 (w,I)Carbon 247 (w,I)Swan band (C2) 467 (m,II), 517 (s,II); 550 (m,II)CN band 388 (s)Hydrogen 656.4 (s,alpha-hydrogen)Magnesium 279.69 (m,I,II); 285.36 (w,I); 293.50 (w,I,II); 382.93 (s,I); 518.30 (s,I)Nitrogen 500 (s,II); 746.8 (w,I)Oxygen 777 (w,triplet)Sodium 588–589 (s,doublet)

w: weak; m: medium; s: strong.

ple #

rTra

Fig. 5. Sam

atios may be assigned to identify the composition of kidney stones.hirty determinations were carried out for each reference andeal sample, each determination being the average of thirty shotsnd discrepancy values are evaluated by the simple Grubbs’ test

Fig. 6. Sample #

1 spectra.

2 spectra.

(p = 0.05). A simple application was designed for data treatment;this software application allows getting to know the intensity valuein both emission lines and molecular bands. In addition, it auto-matically calculates all ratios, their average in 10 shots, standard

J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360 357

ple #

dsl2alads

Fig. 7. Sam

eviation and %RSD, while all ratios are checked by means of theimple Grubbs’ test. Selected emission lines were under the fol-owing experimental conditions: calcium at 318.0 nm, carbon at47.9 nm, hydrogen at 656.2 nm, magnesium at 383.8 nm, nitrogent 744.4 nm, oxygen at 777.4 nm and phosphor at 253.6 nm; these

ines are free from spectral interference. Echelle spectrographllowed us getting a signal-ratio-free sample matrix, laser andetector fluctuations, since this spectrograph is able to show highpectral windows (from 200 to 975 nm) with very high resolution

Fig. 8. Sample #

3 spectra.

(0.1 ´̊A), detecting all elements and bands simultaneously withoutoverlapping.

Calculated ratios were: (i) H/C ratio is the most important ratiosince it allows identifying [11] and differentiating [15] organic com-pounds; both uric acid and cystine are identified by this ratio, and

it also may be used to identify calcium oxalate; (ii) O/N ratio underatmospheric conditions: not very useful in this study; (iii) Ca/C ratioindicates calcium sales, differentiating between calcium oxalateand calcium phosphate, since CaP values must be higher than CaOx

4 spectra.

3 / Tala

vvtabc

3

3c

apf

stc(6cCi(34(M5ST

ael(iw

pabltirt

ccfptd

aTs(pibct

at 1000 ns.Now C:Ca:H:Mg:N:O:P ratios of reference samples are calculated

by using chemical standards. These values are used for calculat-ing the critical values of these ratios, which shall in turn be used

58 J. Anzano, R.-J. Lasheras

alues; (iv) Mg/C ratio indicates magnesium sales, whenever thisalue is high; (v) Ca/H ratio identifies calcium sales and differen-iates between them; (vi) Mg/H ratio identifies magnesium sales;nd (vi) P/(P + C) ratio indicates the presence of phosphates, thus,eing possible to differentiate between calcium oxalates and cal-ium phosphate.

. Results and discussion

.1. Strategy A: �-LIBS system with compact detector (CCD) andorrelation method

The crater diameter used by �-LIBS was approximately 0.18 mmnd irradiance was 7.1 × 109 W cm−2. In spite of the low laser energyulse used, high irradiance was obtained in this system due toocalization by means of a microscope.

The LIBS spectra used as reference to prepare the library arehown in Fig. 4. Cystine and uric acid spectra (Fig 4A and B, respec-ively) are very similar. The most important emission lines for bothompounds are: one strong line appears approximately at 500 nmmaybe atmospheric nitrogen N II [10,18]); alpha-hydrogen (H�) at56 nm and oxygen (O I) at 777 nm. The molecular bands found are:arbon Swan bands from carbon dimmer (C2) at 517 and 550 nm andN bands at 388 nm. Average spectra of cystine and uric acid were

ntroduced as reference in the LIBS library. The spectrum of CaOxFig. 4C) shows the main emission lines of calcium at 317.82; 373.6;93.26 and 396.69 nm; and some other weak lines at 430.18, 443.41,45.45, 526.53 and 559.23 nm. Emission lines of N II (500 nm), O I777 nm) and H� (656.3 nm) are also shown. Magnesium lines from

gP (Fig. 4D) were found at 279.69, 285.36, 293.50, 383.27 and17.97 nm. The CaP spectrum (Fig. 4E) looks like CaOx, althoughwan bands appear at 517 and 550 nm in the CaOx spectrum. Inable 1 the emission lines and bands are indicated.

All average spectra were introduced in the spectra librarynd, subsequently, real samples were analyzed. The previouslyxplained methodology was applied and its linear and rank corre-ation coefficients were calculated by using two spectral windows:a) from 200 to 400 nm, and (b) from 200 to 800 nm. Atmosphericnterferences (N II and O II) may be eliminated by using a width

indow.Samples were analyzed by IR spectroscopy. The first real sam-

le spectra (M1) show the presence of calcium as majority elementnd a weak presence of magnesium (Fig. 5). This sample may thene CaOx and/or CaP, although this stone may be composed of a

ittle amount of MgP. Linear correlation coefficients calculated forhe narrowest spectral window indicate that sample compositions CaOx (r = 0.9632, 0.9929, 0.9858, 0.9703, 0.9100 and 0.9732). Theank correlation coefficient is not as accurate as the linear correla-ion one.

The second real sample spectra (M2) show the absence of cal-ium and magnesium and the sodium line is weak (Fig. 6); onlyalcium and sodium traces were present in some spectrum. There-ore, this sample may be UA or CY. The linear correlation coefficientoints to CY (0.977, 0.9273, 0.9937 and 0.9718) using the width spec-ral window. Worse results were again obtained by using all spectralata and rank correlation.

In the third sample spectra (M3) calcium and magnesium linesppear (Fig. 7). Then, correlation coefficients must be calculated.hese coefficients point to a CaP and MgP mix. Finally, the lastample spectra (M4) do not present calcium and magnesium linesFig. 8), although calcium is detected on external surface, due to

recipitation of calcium sales on the original stone. In this case,

t is important to identify the composition of stone internal-coreecause kidney-stone formation is due to this compound. Then, thisore may be UA or CY, the linear coefficient (200–400 nm) pointingo UA (r = 0.9925).

nta 79 (2009) 352–360

3.2. Strategy B: Echelle spectrograph and ratio method

The crater diameter used in this second experimental design was0.4 mm; thus, irradiance was 5.2 × 109 W cm−2. It was lower thanthe �-LIBS system, despite laser energy was thirteen times higherand pulse duration was reduced to a half in this second system. Thiswas due to the focalization system of the microscope in the �-LIBSsystem.

Signal/noise (S/N) value was studied for delay-time optimiza-tion, and carbon and hydrogen emission lines were analyzed. Delaytime was optimized between 100 and 2000 ns by using the carbonand hydrogen lines at 247 and 656 nm, respectively. S/N variationsdepending on delay time are shown in Fig. 9A and B for carbon andhydrogen line, respectively. A high intensity value for carbon can beobserved at 247.9 nm (Fig. 9A) at low delay times (<500 ns). How-ever, S/N ratios were very low at these times, even below detectionlimit (S/N < 3), due to high noise (background). Optimal delay-timefor carbon was 1000 ns. The behaviour of intensity and S/N val-ues for hydrogen (Fig. 9B) were similar and optimal delay timewas 500 ns; this may happen since lifetime of hydrogen emissionis shorter. On the other land, hydrogen S/N values were higherthan carbon’s values. Therefore, the delay time used as optimalwas 1000 ns. Integration time was also optimized, being optimal

Fig. 9. Variation of S/N depends on delay time.

J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360 359

Table 2C:Ca:H:Mg:N:O:P ratios calculated using chemical standards.

UA CY CaOx CaP MgP

H/C1 5.86 ± 0.57 11.4 ± 1.3 6.9 ± 0.9 23.9 ± 3.4 33.3 ± 6.6O/N 3.3 ± 0.2 3.2 ± 0.2 2.6 ± 0.3 3.5 ± 0.2 3.4 ± 0.2Ca/C1 0.47 ± 0.06 0.88 ± 0.25 42.0 ± 12.0 376 ± 52 4.4 ± 0.6Mg/C1 1.45 ± 0.19 1.4 ± 0.3 3.6 ± 0.5 11.5 ± 1.7 70.5 ± 8.8CMP

n

tt3lpi

(psDretogoTwae

a/H 0.084 ± 0.015 0.050 ± 0.003g/H 0.23 ± 0.03 0.11 ± 0.02

/(P + C1) 0.11 ± 0.01 0.16 ± 0.04

= 30, p = 0.05.

o identify kidney composition. The emission lines chosen underhese experimental conditions for C, Ca, H, Mg, N, O and P are 247.9,18.0, 656.2, 383.8, 744.4, 777.4 and 253.6 nm, respectively. Theseines are free from spectral interference. Due to the high resolutionrovided by Echelle spectrograph, all selected lines are free from

nterferences from studied elements.C:Ca:H:Mg:N:O:P ratios calculated by using chemical standards

UA, CY, CaOx, CaP and MgP) are shown in Table 2 (n = 30 and= 0.05) and their spectra in Fig. 10. The phosphorus emission line

howed the lowest sensitivity at 253.6 nm in studied elements.etection limit was calculated as the amount with a signal/noise

atio ≥3; thus, phosphorus detection limit was 2.3% under thesexperimental conditions. This limit was enough for the identifica-ion of urinary calculus; only majority compounds were the aimf the present study. Compounds without either carbon or hydro-en or nitrogen in their chemical formulas (e.g., CaP and MgP) or

nly one of these elements would produce ratios of H/C, O/N, etc.hese ratios shall be produced since emission at the correspondingavelengths comes from atmospheric sources of C, H, N (CO2, H2O

nd N2) and the most important source may be impurities from ref-rence materials and samples. Something similar occurs with the

Fig. 10. Spectra obtained fro

5.8 ± 1.2 16.2 ± 1.2 0.17 ± 0.030.51 ± 0.05 0.51 ± 0.06 2.6 ± 0.40.46 ± 0.04 0.88 ± 0.02 0.85 ± 0.02

presence of Ca and Mg in UA and CY patterns, Ca in magnesiumphosphate and Mg in calcium salts, whose presence are impuri-ties in patterns. These impurities do not affect analysis results,since only majority elements are indicators of the composition ofrenal calculus, which may also contain remains of other compoundsdifferent than the majority one or that which originates calculusformation (once calculus has been formed, other compounds maysettle on this nucleus).

UA and CY ratios are similar, only the H/C1 ratio can be used todifferentiate between them. These compounds can differ of calciumand magnesium sales due to higher Ca and Mg ratios. CaOx differsfrom CaP since the Ca/C1 ratio of CaOx is lower than that of CaP dueto the presence of carbon in CaOx; besides, P/(P + C1) ratio of CaP ishigher. All ratios are used to assign the critical values which shallbe used to analyze real samples. These critical values are shown inTable 3.

Some kidney stones were analyzed by using these ratios andobtained values are shown in Table 4. Sample #1 could not be cal-cium and magnesium sales since Ca/C1 and Mg/C1 ratios are below2 (1.6 and 1.9, respectively) and Ca/H and Mg/H are also below 0.5.Then, it may be CY or UA: the H/C1 ratio (6.5) points to CY. Sample #2

m chemical standards.

360 J. Anzano, R.-J. Lasheras / Talanta 79 (2009) 352–360

Table 3Critical values used for the identification of kidney stones.

Uric acid Cystine Calcium oxalate Calcium phosphate Magnesium phosphate

H/C1 <6 >6 >6 >6 >6Ca/C1 <2 <2 60 > r > 10 >60Mg/C1 <2 <2 >20Ca/H <0.5 <0.5 8 > r > 4 >8Mg/H <0.5 <0.5P/(P + C1) <0.5 <0.5 <0.6

Italicized text: the most important values for its identification.

Table 4Results of analyzed samples.

Sample #1 Sample #2 Sample #3

H/C1 6.5 3.4 8.5Ca/C1 1.6 1.5 80Mg/C1 1.9 2.0 4.6Ca/H 0.26 0.47 9.5MP

i6cctsI

4

owvctItoic(siilAimmg

[

g/H 0.30 0.60 0.55/(P + C1) 0.36 0.40 0.70

s similar to Sample #1; again it may be CY or UA. As H/C1 is below(3.4), then this sample is UA. Calcium ratios in Sample #3 point to

alcium sales; therefore, it may be CaOx or CaP. The P/(P + C1) ratioonfirms the presence of phosphates as majority salt in the sample;herefore, it is CaP. H/C1 and Ca/H ratios also confirm this compo-ition. All LIBS identifications of kidney stones were confirmed byR spectroscopy.

. Conclusions

In this work two LIBS instrument designs have been devel-ped for identification of urinary calculi: (a) a micro-LIBS systemith a compact detector, CCD camera, (�-LIBS); and (b) a con-

entional LIBS system with an Echelle spectrograph and an ICCDamera. One data-treatment method was studied for each sys-em and real samples of kidney stones – previously analyzed byR spectroscopy – were used for reliable evaluation of two iden-ification strategies. In both instrumentations, visual examinationf the spectra was enough to distinguish between organic (UA, fornstance) and non-organic compounds (CaP, for instance). Statisticalorrelation analysis using linear (parametric) and non-parametricrank) correlation was applied in the LIBS system. Besides, twopectral windows (200–400 and 200–850 nm) were studied due tontense interferences – atmospheric nitrogen and oxygen – appear-ng at high wavelengths. The best results were obtained by usinginear correlation and spectral windows between 200 and 400 nm.

nother identification system was carried out by using different

nstrumentation and a method based in intensity-peak ratios. Thisethod reached successful results in all analyzed samples. Bothethods and instrumentations have been proven as reliable strate-

ies for urinary-calculi identification, although the method based[

>1>0.6 >0.6

on intensity-peak ratios may provide better analytical informationabout the analyzed sample.

Acknowledgements

This work was supported by the Environment Ministry of theSpanish National Government through projects #241/2005/2B-2.7, #634/2006/2-2.7 and #344/2007/3-2.7. We would also like togive special thanks to the Department of Science, Technology andUniversity of the Aragon Regional Government for financing ourresearch group.

References

[1] M.A. Aibar, A.P. Gutierrez, M.P. Rodrigo, K. Laborda, A.B. Hernández, G. Blaso,Acta Urol. Esp. 28 (2004) 661–665.

[2] M. Volmer, J.C.M. De Vries, H.M.J. Goldschmidt, Clin. Chem. 47 (2001)1287–1296.

[3] M. Volmer, A. Block, B.G. Wolthers, A.J. de Ruiter, D.A. Doornbos, W. van der Slik,Clin. Chem. 39 (1993) 948–954.

[4] M. Volmer, B.G. Wolthers, H.J. Meting, T.H.Y. de Haan, P.M. Coenegracht, W. vander Slik, Clin. Chem. 40 (1994) 1692–1697.

[5] L. St-Onge, E. Kwong, M. Sabsabi, E.H.B. Vadas, J. Pharm. Biomed. Anal. 36 (2004)277–284.

[6] C. Samuels, F.C. DeLucia, K.L. McNesby, A.W. Miziolek, Appl. Opt. 42 (2003)6205–6209.

[7] P.B. Dixon, D.W. Hahn, Anal. Chem. 77 (2005) 631–638.[8] O. Samek, H.H. Telle, D.C.S. Beddows, BMC Oral Health 1 (2001) 1–9.[9] B. Gornushkin, A. Ruiz-Medina, J.M. Anzano, B.W. Smith, J.D. Winefordner, J.

Anal. Atom. Spectrom. 15 (2000) 581–586.[10] J. Anzano, M. Casanova, M.S. Bermúdez, R.J. Lasheras, Polym. Test 25 (2006)

623–627.[11] M. Tran, Q. Sun, B.W. Smith, J.D. Winefordner, J. Anal. Atom. Spectrom. 16 (2001)

628–632.12] F. Ferioli, P. Puzinauskas, V. Paulius, S.G. Buckley, Appl. Spectrosc. 57 (2003)

1183–1189.[13] S. Kaski, H. Häkkänen, J. Korppi-Tommola, J. Anal. Atom. Spectrom. 19 (2004)

474–478.[14] L. St-Onge, E. Kwong, M. Sabsabi, E.B. Vadas, Spectrochim. Acta Part B: Atom.

Spectrom. 57 (2002) 1131–1140.[15] C. López-Moreno, S. Palanco, J.J. Laserna, F. DeLucia, A.W. Miziolek, J. Rose, R.A.

Walters, A.I. Whitehouse, J. Anal. Atom. Spectrom. 21 (2006) 55–60.[16] G. Asimellis, A. Giannoudakos, M. Kompitsas, Spectrochim. Acta Part B: Atom.

Spectrom. 61 (2006) 1253–1259.[17] R. Sattmann, I. Monch, H. Krause, R. Noll, S. Couris, A. Hatziapostolou, C. Mavro-

manolakis, E. Fotakis, R. Larrauri, Miguel, Appl. Spectrosc. 52 (1998) 456–461.[18] J.M. Anzano, I.B. Gornushkin, B.W. Smith, J.D. Winefordner, Polym. Eng. Sci. 40

(2000) 2423–2429.[19] X. Fang, S.R. Ahmad, M. Mayo, S. Iqbal, Lasers Med. Sci. 20 (2005) 132–137.20] V.K. Singh, A.K. Rai, P.K. Rai, P.K. Jindal, Lasers Med. Sci. (2008),

doi:10.1007/s10103-008-0635-2.