Journal of Cranio-Maxillofacial Surgery (2008) 36, 443e449

� 2008 European Association for Cranio-Maxillofacial Surgery

doi:10.1016/j.jcms.2008.04.003, available online at http://www.sciencedirect.com

Dimensional error in selective laser sintering and 3D-printingof models for craniomaxillary anatomy reconstruction*

Daniela Nascimento SILVA, MSc, PhD1,*, Marılia GERHARDT DE OLIVEIRA, PhD1,*,

Eduardo MEURER, PhD2,*, Maria Ines MEURER, PhD3,*, Jorge Vicente LOPES DA SILVA, MSc4,*,

Ailton SANTA-BARBARA, MSc4,*

1Department of Surgery (Head: Prof. Dr. Helena Willhelm de Oliveira), Pontifıcia Universidade Catolicado Rio Grande do Sul (PUCRS), Porto Alegre, RS, Brazil; 2Dentistry School, Universidade do Sul de SantaCatarina (UNISUL), Tubar~ao, SC, Brazil; 3Department of Pathology (Head: Prof. Dr. Alcıbia Helena deAzevedo Maria), Universidade Federal de Santa Catarina (UFSC), Florianopolis, SC, Brazil;4Centro de Pesquisas Renato Archer (CenPRA), Campinas, SP, Brazil

SUMMARY. Background: Selective laser sintering (SLS) and three-dimensional printing (3DP�) are rapidprototyping (RP) techniques to fabricate prototypes from biomedical images. To be used in maxillofacial surgery,these models must accurately reproduce the craniofacial skeleton. Purpose: To analyze the capacity of SLSand 3DP� models to reproduce craniomaxillary anatomy and their dimensional error. Material: Dry skull,helical computed-tomography images, SLS and 3DP� prototypes, and electronic calliper. Methods: Tomo-graphic images of a dry skull were manipulated with the InVesalius biomedical software. Prototypes werefabricated using SLS and 3DP� techniques. Ten linear measurements were made on the models andcompared with corresponding dry skull measurements (criterion standard) carried out with an electroniccalliper. Results: We observed a dimensional error of 2.10 and 2.67% for SLS and 3DP� models, respectively.The models satisfactorily reproduced anatomic details, except for thin bones, small foramina and acute bone pro-jections. The SLS prototypes showed greater dimensional precision and reproduced craniomaxillary anatomymore accurately than the 3DP� models. Conclusion: Both SLS and 3DP� models provided acceptableprecision and may be useful aids in most maxillofacial surgeries. � 2008 European Association for Cranio-Maxillofacial Surgery

Keywords: craniofacial, precision, rapid prototyping

INTRODUCTION

Biomedical prototypes or models have been largely usedin maxillofacial surgery, as an aid to diagnosis and treat-ment planning. However, the quality and precision of dif-ferent rapid prototyping (RP) systems have not beendefinitely established (Berry et al., 1997; Schneideret al., 2002; Meurer et al., 2003).

New RP technologies, such as selective laser sintering(SLS) and three-dimensional printing (3DP�) havea lower cost and shorter manufacturing times than tradi-tional stereolithography (SL).

The SLS technique uses a CO2 laser beam to selec-tively fabricate models in consecutive layers. First, thelaser beam scans over a thin layer of powder previouslydeposited on the build tray and levelled with a roller. Thelaser beam heats the powder particles and fuses them toform a solid layer, and then moves along the X and Yaxes to design the structures according to ComputerAided Design (CAD) data. After the first layer fuses,

*Financial support: Conselho Nacional de Desenvolvimento Cientı-

fico e Tecnologico (CNPq).*

These authors contributed equally.

443

the build tray moves down, and a new layer of powderis deposited and sintered. When the manufacturing pro-cess is complete, the prototype is removed from thetray, and the surrounding unsintered powder is dustedoff. The prototype surface is finished by sandblasting.The SLS prototype is opaque, and its surface is abrasiveand porous (Berry et al., 1997; Meurer et al., 2003).

The 3DP� system uses print heads to selectively dis-perse a binder onto the powder layers. This technologyhas a lower cost than similar techniques. First, a thinlayer of powder is spread over a tray using a roller similarto the one used in the SLS system. The print head scansover the powder tray and delivers continuous jets of a so-lution that binds the powder particles as it touches them.No support structures are required while the prototype isfabricated because the surrounding powder supports theunconnected parts. When the process is complete, thesurrounding powder is aspirated. In the finishing process,the prototype surfaces are infiltrated with a cyanoacry-late-based material (Ashley, 1991; Sachs et al., 1992a).

This study evaluated the dimensional error and the re-producibility of anatomical details in prototypes producedusing the SLS and 3DP� technologies (powder-basedaddition systems) in comparison with a dry human skull

444 Journal of Cranio-Maxillofacial Surgery

(criterion standard) using the same protocol for acquisitionand manipulation of computed-tomography (CT) images.

METHODS

A human dry skull was positioned with the Frankfurtplane parallel to the scan plane of a helical CT unit (So-matom Plus 4, Siemens, Munich, Germany) at HospitalSao Lucas, Pontifıcia Universidade Catolica do RioGrande do Sul (PUCRS), Porto Alegre, Brazil. CT im-ages were acquired using the following parameters:0� gantry tilt; axial slices (193 slices); 1 mm slice thick-ness; 1.5 pitch; field of view (FOV)¼ 20.8 cm; matrix512� 512; 120 kVp; and 130 mA.

Skull volume was reconstructed with a slice thicknessof 0.5 mm. CT data were recorded in a Digital ImagingCommunications in Medicine (DICOM) 3.0 CD-R. TheInVesalius software (CenPRA, Campinas, Brazil) wasused to segment images at a 400e3300 threshold; fileswere converted to Standart Template Library (STL) for-mat and sent to the Renato Archer Research Centre (Cen-PRA, Campinas, Brazil) for the fabrication of prototypes.

The unit used for SLS prototyping was a Sinteristation2000 (DTM, USA), and the material was a thin polyamidepowder (PA 2200, EOS, Munich, Germany). The proto-types were finished with sandblasting. Prototype fabrica-tion time was 15 h, and approximate cost was $600.

A ZPrinter� 310 System (MIT, MA, USA) unit wasused for the production of the 3DP� prototypes, andthe materials were plaster powder (zp�102, Z Corpora-tion, Burlington, USA) and a water-based binder. Theprototypes were finished by application of a cyanoacry-late-based infiltration material (Z-Bond100, Z Corpora-tion, Burlington, USA) to their surfaces. Prototypefabrication time was 4 h, and approximate cost was $430.

Fig. 1 shows the sequence of steps for prototype fab-rication.

Fig. 1 e General process of SLS a

Linear measurements (Table 1 and Fig. 2) were madewith a Starrett� electronic calliper (Starrett Ind. e Com.Ltda, Sao Paulo, Brazil). Data were analyzed using de-scriptive statistics, and comparisons were made with theStudent t test for paired samples. For each linear measure-ment, dimensional error was calculated as the absolute dif-ference (mm) between the values obtained from theprototypes and those from the dry skull. Relative differ-ences (%) were calculated as the absolute difference di-vided by the dry skull value� 100, in accordance withstudies conducted by Choi et al. (2002) and Chang et al.(2003). Each measurement of prototypes and dry skullwas repeated 20 times by the same observer, and resultswere used for the subsequent comparison of mean values.

Mean absolute difference ðmmÞ¼ prototype value� dry skull value

Mean relative difference ð%Þ

¼ ðprototype value � dry skull valueÞ � 100

dry skull value

RESULTS

Fig. 3 shows that the external measurements of the pro-totypes were greater than those of the dry skull, exceptfor length palatal (LP). Internal prototype measurementvalues were lower than dry skull values, except for thezygomaticofrontal (ZFeZF) and length of internal cra-nium (LIC) of the SLS prototype. All differences werestatistically significant at p # 0.05.

Figs. 4 and 5 show the mean differences in linear mea-surements of the prototypes and of the dry skull. These

nd 3DP� model production.

Table 1 e Landmarks and linear measurements

Landmarks and measurements Definition

LandmarksAP e aperture piriformis Point at the lateral

margin of the aperturepiriformis (bilateral)

Ba e basion The median point of theanterior margin of theforamen magnum

FC e frontal crest Tip of the bony frontalcrest

IOC e internal occipital crest The anterior-most pointof the internaloccipital crest

ANS e anterior nasal spine Tip of the bonyanterior nasal spine

PNS e posterior nasal spine The median point on thedorsal limit of palate

EF e external frontal The anterior-most pointof the frontalin the median plane

LM e lateral foramen magnum Point at the lateral-mostmargin of the foramenmagnum (bilateral)

FO e foramen ovale Point at the medialmargin of the foramenovale (bilateral)

ZF Point at the medialmargin of thezygomaticofrontalsuture (bilateral)

EO e external occipital The median point ofthe anterior marginof the occipital

Op e opisthion The median point of theposterior margin ofthe foramen magnum

T e tuberosity Point at the lateralmargin of the maxillarytuberosity (bilateral)

Zy e zygion Point at the lateral-mostborder of the centreof the zygomaticarch (bilateral)

External measurementsLEC e length of external cranium Distance between EF

and EOLP e length of palatal Distance between ANS

and PNSFOeFO Distance between left

and right FOBW e bizygomatic width Distance between left

and right ZyMW e maxillary width Distance between left

and right T

Internal measurementsLFM e length of foramen magnum Distance between Ba

and OpLIC e length of internal cranium Distance between FC

and IOCZFeZF Distance between left

and right ZFAPW e aperture piriformis width Distance between left

and right APFMW e foramen magnum width Distance between left

and right LFM

Dimensional error in selective laser sintering and 3D-printing models 445

data were used to calculate the mean relative differencesof all linear measurements. The SLS prototype hada mean dimensional error of 0.89 mm (2.10%), and the3DP� prototype, of 1.07 mm (2.67%).

DISCUSSION

Most systems used to fabricate biomedical models pro-vide satisfactory accuracy. However, the shape, dimen-sions and anatomic details of prototypes may beaffected by errors at any phase of the RP process, suchas CT image acquisition, image manipulation with bio-medical software, or fabrication and finishing (Barkeret al., 1994; Choi et al., 2002; Schneider et al., 2002;Tang et al., 2003).

Some parameters should be carefully analyzed to en-sure accuracy when using the SLS technology: slicethickness when the CAD model is resliced, diameterand angle of the CO2 laser beam, type and size of powderparticles, and direction of fabrication (Tang et al., 2003).

In the 3DP� system, the printing mechanism, the typeand quality of the materials used in the fabrication of theprototypes, and the absorption properties of the powderwhen in contact with the binder and infiltration materialare parameters that should be controlled to obtain a reli-able final product (Sachs et al., 1992a; Chang et al.,2003).

The analysis of external measurements (Fig. 3)showed that the length of external cranium (LEC), bizy-gomatic width (BW), FOeFO and maxillary width(MW) values of the SLS and 3DP� prototypes weregreater than those of the dry skull. Only the LP valuewas lower. This may have occurred because the referencepoints for this measurement (anterior nasal spine [ANS]and posterior nasal spine [PNS]) are in areas of acutebone projections, and are, thus, more susceptible to thepartial volume effect, which attenuates projections andconsequently reduces the LP dimension. Choi et al.(2002) reported similar results in their analysis of theSL technique. Conversely, all internal measurementvalues of the 3DP� prototype were lower than thoseof the dry skull (Fig. 3). The same was found for theSLS prototype, except for LIC and ZFeZF, whose ana-tomic landmarks are also located in areas of acute boneprojections (frontal crest [FC] in the LIC dimension) orin bone sutures (zygomaticofrontal suture, in the ZFeZF dimension), regions that are greatly affected by thepartial volume effect (Choi et al., 2002).

The inverse correlation between external and internaldimensions may be explained by the dumb-bell effectdescribed by Choi et al. (2002), in which the increasein external dimensions and a simultaneous decrease ininternal dimensions indicate that the prototypes havegreater dimensions than the original skull, and that theselected threshold may have been too low. Therefore,accuracy is dependent primarily on the choice of a scan-ning protocol and on data segmentation, especially thedetermination of threshold.

Table 2 summarizes relevant studies about the accu-racy of 3D CT imaging and RP techniques in the repro-duction of craniofacial anatomy.

Fig. 2 e Linear measurements.

446 Journal of Cranio-Maxillofacial Surgery

Results of this study showed a dimensional error of2.10% for the SLS prototype in comparison with thedry skull, a value that is greater than that reported byBerry et al. (1997), who found a 0.64% variation. How-ever, those authors compared the prototype with the 3DCT image, whose dimensions might have been differentfrom those of the original skull, and did not adopt a crite-rion standard. Choi et al. (2002) investigated the accu-racy of SL prototypes produced from CT data of a dry

0

20

40

60

80

100

120

140

160

180

200

LEC BW LP FO-FO MW

External

Linear m

Mean

(m

m)

Dry skull SLS m

Fig. 3 e Mean and standard deviation (SD) of the linear dimensions of dry sBW¼ 123.11 (^0.20), LP¼ 52.04 (^0.27), FOeFO¼ 48.41 (^0.69), MWFMW¼ 27.16 (^0.11), ZFeZF¼ 99.36 (^0.22), APW¼ 26.25 (^0.15). S(^0.21), FOeFO¼ 50.32 (^0.13), MW¼ 74.53 (^0.31), LIC¼ 155.16 (^ZF¼ 99.55 (^0.30), APW¼ 25.26 (^0.14). 3DP model: LEC¼ 186.76 (^(^0.16), MW¼ 74.73 (^0.16), LIC¼ 153.86 (^0.12), LFM¼ 32.52 (^0(^0.40).

skull. Their results showed a mean absolute differenceof 0.62 ^ 0.35 mm (0.56 ^ 0.39%). Chang et al.(2003) evaluated the accuracy of the 3DP� techniquein RP of three types of bone defects: unilateral maxillec-tomy, maxillectomy, and orbitomaxillectomy in freshcadaver skulls. The defects simulated resections ofa tumour in the maxillary sinus. Their results showedthat mean error was lower than 2 mm. According toWaitzman et al. (1992), the dimensional error of 3D

LIC LFM FMW ZF-ZF APW

Internal

easurements

odel 3DP™ model

kull, SLS and 3DP� models. Dry skull: LEC¼ 184.43 (^0.26),¼ 73.40 (^0.14), LIC¼ 154.62 (^0.36), LFM¼ 34.61 (^0.07),

LS model: LEC¼ 187.37 (^0.19), BW¼ 125.22 (^0.14), LP¼ 51.460.19), LFM¼ 33.49 (^0.07), FMW¼ 26.17 (^0.16), ZFe

0.14), BW¼ 125.28 (^0.10), LP¼ 51.66 (^0.12), FOeFO¼ 50.51.07), FMW¼ 25.29 (^0.18), ZFeZF¼ 98.60 (^0.48), APW¼ 24.54

-2,5-2,0-1,5-1,0-0,50,00,51,01,52,02,53,0

LEC BW LP FO-FO MW LIC LFM FMW ZF-ZF APW

External InternalLinear measurements

Ab

so

lu

te d

ifferen

ces M

ean

(m

m)

SLS model - Dry skull 3DP™ model - Dry skull

Fig. 4 e Comparison of linear measurements of dry skull, SLS and 3DP� models: absolute differences’ means.

Dimensional error in selective laser sintering and 3D-printing models 447

CT is about 0.9%, whereas Asaumi et al. (2001) founda variation of 2.16%.

In this study, the 3DP� prototypes had a mean errorof 2.67%, a value slightly lower than the one reportedby Chang et al. (2003). It is important to point out thatthose authors used fresh cadaver skulls with soft tissues.Additional studies should evaluate the effect of these tis-sues on 3D CT imaging and biomedical prototypes.

Our results demonstrated that SLS prototypes repro-duced the craniomaxillary dimensions with a slightlygreater accuracy than the 3DP� prototypes. One factorthat may partly explain the smaller dimensions of theSLS prototypes is the superficial wear caused by sand-blasting (Berry et al., 1997; Meurer et al., 2003). Addi-tionally, the dimensions of 3DP� prototypes may havebeen greater because of cyanoacrylate infiltration (Sachset al., 1992a).

The unused powder that surrounds the prototype in theSLS system cannot be reused. Because of the high cost ofthe material, several parts are fabricated simultaneously.This may explain the long fabrication time for the SLStechnique (16 h), very close to the time required for fab-rication with the SL system (D’Urso et al., 1998; Choiand Samavedam, 2002; Sannomiya and Kishi, 2002;Mazzonetto et al., 2002). The powder remaining in the3DP� system may be reused, and the parts may be fab-

0

1

2

3

4

5

6

7

8

LEC BW LP FO-FO MW

External Linear me

Relative d

ifferen

ces M

ean

(%

)

SLS model - Dry skull

Fig. 5 e Comparison of linear measurements of dry skull,

ricated individually, which substantially reduces proto-type fabrication time (4 h) (Sachs et al., 1992a,b;Mazzonetto et al., 2002). Therefore, the 3DP� techniquehas a lower final cost than the SLS technique, which, inturn, has a lower cost than the SL technique (Sachs et al.,1992a,b; Stoker et al., 1992; Mazzonetto et al., 2002;Meurer et al., 2003).

The dimensional error of the SLS and 3DP� proto-types was within acceptable values. Asaumi et al.(2001) suggest that dimensional changes may not affectthe success of surgical applications if such changes arewithin a 2% variation. However, it is necessary to iden-tify the necessary level of RP accuracy for specific surgi-cal procedures in maxillofacial surgery and to adapt theuse of prototypes to each procedure.

The dimensional accuracy found here may be satisfac-tory when the prototypes are used for communicationwith patients, diagnosis, or presurgical planning, particu-larly for more complex surgeries, such as correction oflarge bone displacement due to trauma and severe facialdeformities (D’Urso et al., 1998; Santler et al., 1998).Chiarini et al. (2004) used RP to plan cranioplasties.For the same purpose, Rotaru et al. (2006) simulatedthe reconstruction of a large cranial defect using SLS pro-totypes. Recently, Paeng et al. (2007) demonstrated theeffectiveness of the preoperative simulation of distraction

LIC LFM FMW ZF-ZF APW

Internalasurements

3DP™ model - Dry skull

SLS and 3DP� models: relative differences’ means.

Table 2 e Comparison with results of other studies on 3D CT or RP eskull differences

Authors Comparison Meandifference (%)

This study SLSedry skull 2.103DP�edry skull 2.67

Waitzman et al.(1992)

3D CTedry skull 0.9 (0.1e3.0)

Ono et al. (1994) SLedry skull 3

Barker et al.(1994)

SLedry skull 0.6e3.6

Berry et al. (1997) SLSe3D CT 0.64

Asaumi et al.(2001)

3D CTedry skull 2.16SLedry skull 0.63

Choi et al. (2002) 3D CTedry skull 0.65SLedry skull 0.56SLe3D CT 0.62

Chang et al. (2003) 3DP�efresh skull 2.1e4.7

448 Journal of Cranio-Maxillofacial Surgery

osteogenesis using biomedical prototypes in the surgicaltreatment of patients with hemifacial microsomia.

Studies conducted by Mazzonetto et al. (2002) andMeurer et al. (2003) showed that prototypes are particu-larly useful in the surgical treatment of temporomandib-ular joint ankylosis because they make it possible toestablish the depth of osteotomy in the drill used asa guide. However, due to the close anatomic relation ofthis joint with important anatomic structures, the dimen-sional error of prototypes should be taken into accountwhen dimensions are transferred to the patient. Suchprocedure is important to prevent accidents, such as therupture of the maxillary artery.

In the determination of the anatomic margins of a tu-mour, the inaccurate representation of tumours in 3DCT and the dimensional error of prototypes (Santleret al., 1998) should be carefully considered.

Further studies should be conducted to investigatewhether the accuracy of prototypes is also satisfactoryfor other surgical procedures, such as dental implantplacement (Poukens et al., 2002; Sarment et al., 2003).

This study demonstrated that the accuracy of the RPsystems should be evaluated for each of its fabricationstages. Technological advances have contributed signifi-cantly to the improvement of techniques that minimizeprototype dimensional error. The development of newtechniques, such as SLS and 3DP�, may contribute toa reduction of costs of the final product and benefita greater number of patients.

CONCLUSIONS

The SLS and 3DP� prototypes fabricated from CT im-ages reproduce craniomaxillary dimensions with accept-able precision, and may be, therefore, useful in mostmaxillofacial operations. SLS prototypes have a greaterdimensional precision and reproduce anatomical detailsof the craniomaxillary region more accurately than3DP� prototypes.

ACKNOWLEDGEMENTS

Brazilian National Council for Scientific and Technolog-ical Development (Conselho Nacional de Desenvolvi-mento Cientıfico e Tecnologico e CNPq); RenatoArcher Research Center (Centro de Pesquisas RenatoArcher e CenPRA).

References

Asaumi J, Kawai N, Honda Y, Shigehara H, Wakasa T, Kishi K:Comparison of three-dimensional computed tomography with rapidprototype models in the management of coronoid hyperplasia.Dentomaxillofac Radiol 30: 330e335, 2001

Ashley S: Rapid prototyping systems. Mech Eng 113: 34e43,1991

Barker TM, Earwaker WJS, Lisle DA: Accuracy ofstereolithographic models of human anatomy. Australas Radiol38: 106e111, 1994

Berry E, Brown JM, Connell M, Craven CM, Efford ND, Radjenovic A,Smith MA: Preliminary experience with medical applications ofrapid prototyping by selective laser sintering. Med Eng Phys 19:90e96, 1997

Chang PS, Parker TH, Patrick Jr CW, Miller MJ: The accuracy ofstereolithography in planning craniofacial bone replacement. JCraniofac Surg 14: 164e170, 2003

Chiarini L, Figurelli S, Pollastri G, Torcia E, Ferrari F, Albanese M,Nocini PF: Cranioplasty using acrylic material: a new technicalprocedure. J Craniomaxillofac Surg 32: 5e9, 2004

Choi JY, Choi JH, Kim NK, Kim Y, Lee JK, Kim MK, Lee JH,Kim MJ: Analysis of errors in medical rapid prototyping models.Int J Oral Maxillofac Surg 31: 23e32, 2002

Choi SH, Samavedam S: Modelling and optimisation of rapidprototyping. Compu Ind 47: 39e53, 2002

D’Urso PS, Atkinson RL, Lanigan MW, Earwaker WJ, Bruce IJ,Holmes A, Barker TM, Effeney DJ, Thompson RG:Stereolithographic (SL) biomodelling in craniofacial surgery. Br JPlast Surg 51: 522e530, 1998

Mazzonetto R, Moreira RW, Moraes M, Albergaria Barbosa JR,Passeri LA, Spagnoli DB: Uso de modelos estereolitograficos emcirurgia bucomaxilofacial. Rev Assoc Paul Cir Dent 56: 115e118,2002

Meurer E, Oliveira MG, Meurer MI, Silva JV, Barbara AS, Heitz C:Biomodelos de prototipagem rapida em CTBMF. Rev Bras CirPeriod 1: 172e180, 2003

Ono I, Gunji H, Suda K, Kaneko F: Method for preparing an exact-sizemodel using helical volume scan computed tomography. PlastReconstr Surg 93: 1363e1371, 1994

Paeng JY, Lee JH, Lee JH, Kim MJ: Condyle as the point of rotationfor 3-D planning of distraction osteogenesis for hemifacialmicrosomia. J Craniomaxillofac Surg 35: 91e102, 2007

Poukens JM, Haex JK, Verdonck H, Riediger D: Study on the accuracyof stereolithographic surgical guides in dental implantology.Phidias Rapid Prototyp Med 8: 8e9, 2002

Rotaru H, Baciut M, Stan H, Bran S, Chezan H, Iosif A, Tomescu M,Kim SG, Rotaru A, Baciut G: Silicone rubber mould castpolyethylmethacrylateehydroxyapatite plate used for repairinga large skull defect. J Craniomaxillofac Surg 34: 242e246, 2006

Sachs E, Cima M, Bredt J, Curodeau A, Fan T: CAD-casting: directfabrication of ceramic shells and cores by three dimensionalprinting. Manuf Rev 5: 117e126, 1992a

Sachs E, Cima M, Williams P, Brancazio D, Cornie J: Threedimensional printing: rapid tooling and prototypes directly froma CAD model. J Eng Ind 114: 481e488, 1992b

Sannomiya EK, Kishi K: Estereolitografia: aplicac~ao na odontologia:relato de dois casos clınicos. Rev Bras Odontol 59: 203e204,2002

Santler G, Karcher H, Ruda C: Indications and limitations of three-dimensional models in craniomaxillofacial surgery. J OralMaxillofac Surg 26: 11e16, 1998

Sarment DP, Sukovic P, Clinthorne N: Accuracy of implant placementwith a stereolithographic surgical guide. Int J Oral MaxillofacImplants 18: 571e577, 2003

Dimensional error in selective laser sintering and 3D-printing models 449

Schneider J, Decker R, Kalender WA: Accuracy in medical modelling.Phidias Rapid Prototyp Med 8: 5e14, 2002

Stoker GN, Mankovich NJ, Valentino D: Stereolithographic models forsurgical planning. J Oral Maxillofac Surg 50: 466e471, 1992

Tang Y, Loh HT, Fuh JY, Wong YS, Lu L, Ning Y, Wang X: Accuracyanalysis and improvement for direct laser sintering https://dspace.mit.edu/bitstream/1721.1/3898/2/IMST001.pdf; 2003, 2003Available at:

Waitzman AA, Posnick JC, Armstrong DC, Pron GE: Craniofacialskeletal measurements based on computed tomography: part I.Accuracy and reproducibility. Cleft Palate Craniofac J 29: 112e117,1992

Dr. Daniela Nascimento SILVA, MSc, PhDSchool of Dentistry e PUCRSAv. Ipiranga6681 Predio 6, Sala 209CEP 90619-900 Porto AlegreRS, Brazil

Tel./Fax: +55 51 3320 3538E-mail: danitxf@hotmail.com

Paper received 23 April 2007Accepted 28 December 2007