Transcript
TECHNO BYTES Digital dental cast placement in 3-dimensional,full-face reconstruction: A technical evaluation Riccardo Rosati, a Marcio De Menezes, a Alberto Rossetti, a Chiarella Sforza, b and Virgilio F. Ferrario b Milan, Italy Introduction: Several noninvasive methods are used for 3-dimensional (3D) morphologic facial and dentalanalysistoaidpractitionersduringdiagnosisandtreatmentplanning.Integratingdentalandfacialnoninvasive3D reproduction could improve the efficacy of treatment management. Methods: Dental virtual model andsoft-tissue facial morphology were digitally integrated from 11 adults with a 3D stereophotogrammetricimaging system (Vectra, Canfield Scientific, Fairfield, NJ). The digital 3D coordinates of 3 facial landmarks(N, nasion; Ftr, frontotemporale right; Ftl, frontotemporale left) and 3 dental landmarks (I, interincisor; Pr, PI,tips of the mesiovestibular cusps of the right and left first permanent premolars) were then obtained byusing Vectra’s software. Additionally, the coordinates of the same 6 landmarks were digitized directly oneach subject by using a 3D computerized electromagnetic digitizer (in vivo). Seven linear measurementswere made between the occlusal plane (Pr-I-Pl) and the facial landmarks (Ftr-N-Ftl). The accuracy andreliability of the reconstruction were tested by in-vivo measurements and repeated acquisitions. Results: Thegreatestmeanrelativeerrorofmeasurementswassmallerthan1.2%.Nosignificantdifferencesinrepeat-able reproductions were found. Conclusions: Integration of facial stereophotogrammetry acquisition anddental laser scan reproduction is possible with marginal error. (Am J Orthod Dentofacial Orthop2010;138:84-8) C linical assessments based on anthropometricstudies offer a significant change in the processof diagnosis for various syndromes, giving sup-port to plastic and orthognathic surgery, detecting nor-mal and abnormal growth, and providing informationfor planning and evaluating medical procedures andtreatments. 1,2 Studies with three-dimensional (3D) re-constructions reported accurate and reliable techniquesfor facial analyses, aiding clinicians in planning moreeffective treatments. 2-6 Several noninvasive methods such as laser scan-ning, magnetic resonance imaging, contact digitizers,Moire´ stripes, and stereophotogrammetry are usedfor 3D morphologic facial analysis. 7 Stereophotogram-metry is a system that uses 2 or more cameras to re-cord several simultaneous photos from differentviewpoints to obtain a digital 3D image. This tech-nique is a promising method of soft-tissue evaluationthat allows reliable analysis of craniofacial defor-mities, 2,4,8-10 providing fundamental parameters toplan and evaluate treatments and maxillofacialsurgery. 11 Current technology also allows digitizala-tion of dental models in 3D space, and the associationbetween the digital dental cast and the 3D facial im-ages could allow the clinician to analyze the relation-ship between soft tissues and dental arches, avoidingwhen possible the use of x-rays. 10 Rangel et al 10 recently proposed the integration of a digital dental cast into a 3D facial picture. Accordingto the average distance between the matched areas (an-terior teeth) in a patient, the method was reported to bereliable, but larger studies are necessary to verifywhether matching between the digital dental cast andthe 3D facial image could correspond to the correctposition of the whole dental arches.Because the matching between the 3D face and thedigital casts with stereophotogrammetric systems mustuse the anterior teeth as the reference, any displacementor inclination of the digital dental arch would add a po-sitionerrorintheposteriorregion.Thus,somequestionsarise. Is the occlusal plane in the correct position? Arethe posterior teeth in the correct position? Therefore,the aims of this study were to elaborate and validatethe matching between digital dental casts and stereo-photogrammetric images as a noninvasive 3D recon-struction of dentofacial structures in healthy subjects,analyzing distances between the occlusal plane andthe facial landmarks. From the Dipartimento di Morfologia Umana e Scienze Biomediche ‘‘Citta`Studi,’’ Universita` degli Studi di Milano, Milano, Italy. a Postgraduate student. b Professor.The authors report no commercial, proprietary, or financial interest in theproducts or companies described in this article.Reprint requests to: Chiarella Sforza, Dipartimento di Morfologia Umana, viaMangiagalli 31, I-20133 Milano, Italy; e-mail,
[email protected], August 2009; revised and accepted, October 2009.0889-5406/$36.00Copyright 2010 by the American Association of Orthodontists.doi:10.1016/j.ajodo.2009.10.035 84 MATERIAL AND METHODS Healthy subjects were selected to have their maxil-larydentalcastsdigitizedbylaserscanningandtheir3Dfacial soft-tissue acquisitions merged into 1 file. Sevenlinear distances through facial and dental landmarkswere measured and compared between the in-vivo andthe virtual reproductions to quantify the accuracy of the final 3D reconstruction.A group of 11 healthy volunteers, 4 men and 7women, from 20 to 31 years of age, with no historyof craniofacial trauma, no congenital anomalies, andfull maxillary and mandibular dental arches, were se-lected. All procedures were noninvasive and were car-ried out with minimal disturbance to the subjects, whowere previously informed about the procedures andgave written consent to the investigation, accordingto the principles outlined in the Declaration of Hel-sinki.For each subject, a maxillary dental reproductionwas obtained by an irreversible hydrocolloid (Tropical-gin, ZhermackSpA,Badia Polesine, Italy) andcast witha type 3 model dental stone (Elite model, Zhermack SpA). Using a commercial laser scanner (D100, Imetric3D, Courgenay, Switzerland), the dental casts were Fig 1. Facial (stereophotogrammetric 3D reproduction) and dental (laser scan) file-matching steps. Fig 2. Three-dimensional teeth-facial reproduction American Journal of Orthodontics and Dentofacial Orthopedics Rosati et al 85 Volume 138, Number 1 digitized, and the appropriate files were imported intothe stereophotogrammetric software.The soft tissues’ facial morphology was acquiredwitha3Dstereophotogrammetryimagingsystem(Vectra3D, Canfield Scientific, Fairfield, NJ). This is a modular3Dsystemdesignedtocaptureandprocessstereoimages.It consists of 2 pods, including 3 cameras (2 black andwhite, and 1 color) and a projector in each pod.The system records synchronized pairs of 2-dimen-sional images of the subjects. With dedicated software,the information is used towork out the 3D reconstructionsthatsubsequentlycanbeprocessed,analyzed,manipulated,and measured. The reproducibility of stereophotogram-metric technology has been well documented. 1,3,4,8, 9,12,13 Before each acquisition, 3 soft-tissue landmarks (N,nasion; Ftr, frontotemporale right; Ftl, frontotemporaleleft) were marked on the facewith black liquid eye linerfor further analysis. 14 For each subject, 2 sets of 3Dfacial images were obtained: with open lips (with cheek retractors)withvisiblefrontalteeth andwiththeteeth inocclusion and closed lips.To obtain virtual dentofacial reproductions, the 3Dfacial images and the corresponding maxillary digitaldental arch were matched by using Vectra’s softwaretools. The matching process needed 3 steps (Fig 1).1. The digital dental cast was merged with the open-lips facial acquisition, and anterior dental land-marks, present in both acquisitions, were used asfiducial points.2. The image with closed lips was introduced to be re-lated to the open-lips image; the facial soft-tissuelandmarksN,Ftr,andFtlwereusedasfiducialpoints.3. The open-lips acquisition was removed from thefile, obtaining a final digital image (Fig 2), withthe dental cast and the facial reconstruction withclosed lips in the relevant 3D positions.The matching of the 2 surfaces was obtained with 2techniques: an initial approximate alignment was per-formed, overlapping the 3 fiducial points recognizableinbothsurfaces.Theprocedurewasthenimprovedbyus-ing the ‘‘register surface’’ control in the Vectra software.To verify the accuracy of the virtual full reproduc-tion, the 3D coordinates of the 3 facial (N, Ftr, andFtl) and the 3 dental (I, interincisor; Pr and Pl, tips of thevestibular cusps of right and left first permanent firstpremolars) landmarks were obtained directly on eachsubject by using a 3D computerized electromagneticdigitizer(3Draw,Polhemus,Colchester,Vt)withareso-lution of 0.0005 cm per centimeter of range. 15,16 Seven linear measurements were then mathemati-cally computed from the 3D coordinates and calculatedwith Euclidean geometry (Table I). The same linear dis-tances were also obtained from the 3D digital reproduc-tion by using Vectra’s software tools (Fig 3).To investigate the reliability of matching betweenthe images, the merging procedures described above Table I. Differences between distances obtained in vivo and on the virtual reproduction In vivo (mm) Virtual (mm) MAD CI TEM (mm) REM (%)Wilcoxon test P value Distance Mean SD Mean SD (mm) (mm) (mm) Ftr-Pr 86.9 6.5 87 6.1 0.8 0.44 1.16 0.7 0.9 0.66Ftr-Pl 106.7 4.1 106.9 3.4 1.1 0.55 1.65 1 1 0.53N-I 85.8 19.9 86.3 20.1 0.7 0.45 0.95 0.6 0.9 0.051Ftl- Pl 87.4 5.4 87.8 5.5 0.9 0.65 1.15 0.7 1.1 0.29Ftl-Pr 106 3.6 107 3.7 1.2 0.7 1.7 1 1.1 0.03*Ftr-I 101.9 5.8 102.1 5.1 0.9 0.5 1.3 0.8 0.9 0.59Ftl-I 100.8 5.9 101.8 5.9 1.2 0.79 1.61 1 1.2 0.03** P \ 0.05. Fig3. Lineardistancesfromcutaneousanddentalland-marks on the virtual reproduction. 86 Rosati et al American Journal of Orthodontics and Dentofacial Orthopedics July 2010 were done twice, and the distances reported in Table Iwere calculated on each matching.Because the merged images were made with 2photos (closed and open lips, Fig 1, B ), to assess the ac-curacy of Vectra’s matching, a random area comprisingforehead, temples, and nose was selected for eachclosed-lips photograph, and the distance from the re-spective open-lips pictures was evaluated. Ideally, thispart of the face should not move with open lips. Statistical analysis Wilcoxon signed rank tests were made between thedistances computed in vivo and on the virtual reproduc-tion to detect possible systematic errors. P values lessthan 0.05 were considered significant.Mean absolute difference (MAD), technical error of measurement (TEM), and relative error magnitude(REM) were calculated to quantify the precision of ourprotocol. 8 MAD is the average of absolute differencesbetween the values of 2 sets of measurements; TEM orDahlberg’s error 17 was used to evaluate the random er-ror. The REM was obtained by dividing the MAD fora variable by the grand mean for that variable and mul-tiplyingtheresultby100;itrepresentsanestimateofer-ror magnitude relative to the size of the measurement. 8 RESULTS Themeansandstandarddeviationsofthevirtualandin-vivo measurements are reported in Table I. Statisti-cally significant differences were found in 2 distances(Ftl-Pr and Ftl-I), with P values from the Wilcoxon testslessthan0.05.Inalloccasions,themeanabsolutediffer-ences were equal to or less than 1.2 mm.The TEM was between 0.6 and 1 mm. The REM (anestimate of error magnitude relative to the size of themeasurement) ranged between 0.9 and 1.2%.Table II gives the statistical analysis of reliability of the matching procedures. No significant differenceswere found between repeated merging. MAD andTEM values were less than 0.6 mm.The matching process was also accurate: the meandistance between the foreheads in the 2 acquisitions(open lips vs closed lips) was 0.4 mm, ranging between0.04 and 1.1 mm (Table III). DISCUSSION Manydysmorphicsyndromesorcraniofacialanoma-liesthatinvolvedentofacialrelationshipscanbeanalyzedby a combination of soft- and hard-tissue assess-ments. 5,10,16,18 In several clinical applications, virtualreproductions of the morphology can aid practitionersduring diagnosis and planning of medical proceduresand treatments. Currently, x-ray technology allowsreproducing both the 3D morphology of teeth and thesoft-tissue facial structures, but this is invasive becauseof the radiation. 5 Clinicians should consider the risksand benefits to the patient when obtaining craniofacialimages,andcurrentresearchistryingtoreduce unneces-sary radiographic exposure, especially in children. 2,10,16 Inthisstudy,anoninvasiveprotocoltoreproduceden-tofacialfeatureswasdefinedandevaluated.The2virtualfacial reconstructions were well matched, with mean er-rors of 0.4 mm, a value in perfect accord with that re-ported by Rangel et al 10 for patient matching. Bothminimal forehead movements (mainly from mimic Table II. Reproducibility of matching First match Second match Distance (mm) Mean SD Mean SD MAD CI TEM PFtr-Pr 88.26 5.32 88.27 5.31 0.26 0.16 0.36 0.21 0.81Ftr-Pl 106.83 4.49 106.89 4.64 0.53 0.29 0.77 0.46 0.86N-I 98.05 25.01 97.92 24.89 0.31 0.19 0.43 0.25 0.31Ftl- Pl 88.24 5.04 88.33 5.35 0.58 0.27 0.89 0.52 0.86Ftl-Pr 106.67 3.98 106.74 4.21 0.33 0.19 0.47 0.27 0.77Ftr-I 102.43 5.37 102.42 5.16 0.19 0.08 0.3 0.18 0.59Ftl-I 102.11 6.41 101.99 6.35 0.28 0.16 0.4 0.24 0.41 P values from Wilcoxon signed rank test ( P \ 0.05). Table III. Average distance of the 2 matched surfaces Subject Average distance (mm) SD 1 0.35 0.142 0.04 0.163 0.46 0.224 0.1 0.075 0.19 0.166 1.1 0.27 0.16 0.178 0.6 0.159 0.42 0.3210 0.36 0.0911 0.58 0.2Mean 0.40 0.17 American Journal of Orthodontics and Dentofacial Orthopedics Rosati et al 87 Volume 138, Number 1 muscles)anddifferentheadpositions(mainlyrotationsinthe frontal and sagittal planes) can occur between the 2facial acquisitions, making them not perfectly superim-posable. Nonetheless, the modifications were clinicallyimperceptible,andtheerrorcanbeconsiderednegligible.The reliability tests demonstrated good-quality 3Dimage managing, with TEM and MAD less than 0.6mm and no systematic errors. Also, the results showedsatisfactory agreement between the virtual reproduc-tions and the in-vivo acquisitions, at least for the first4 teeth in each hemi-arch. Even though 2 distances(Ftl-Pr and Ftl-I) had statistically significant differ-ences, TEM and MAD were always less than 1.0 and1.2 mm, respectively, with REM up to 1.2%. Overall,these errors werewithin an acceptable range for clinicaland anthropometric contexts. 6,19 These findings couldbe explained by software precision. The high-qualitysurface texture is helpful to recognize the landmarks;this increases matching precision.The main technical complication was due to toothtranslucence allied with saliva biofilm that often re-sulted in unsatisfactory virtual tooth reconstructions.Maintaining a dry tooth surface was an important stepin obtaining satisfactory results. Overall, our protocolwas appropriate to the initial purpose.Further studies are necessary to improve the proto-col and to find a reliable way to associate the virtualmandibular dental arch with the 3D digital facial recon-structions,thusallowingitsdailyuse,mostlyinresearchfacilities.Also,correct dental castpositioning should becontrolled for all dental elements; we tested only inci-sors and first premolars, but positioning errors might in-crease in the posterior part of the dental arch. Anotherlimitofthisprotocol mightbethecost ofthesetechnicalinstruments for a private dental office. CONCLUSIONS Merging digitized dental maxillary arch and 3D ster-eophotogrammetricfacialacquisitionscouldprovideare-liable reproduction of dentofacial relationships, at leastfor the first 4 teeth in each hemi-arch. The method couldbe used to monitor longitudinally the evolution of ortho-donticandorthopedichealingthroughnoninvasiveacqui-sition of dental and facial morphology, thus guidingcliniciansinreal-timemanagementoftreatments.There-ductioninx-rayexposureandtheuseofaglobal3Danal-ysis are probably the main benefits of the method. REFERENCES 1. Aldridge K, Boyadjiev SA, Capone GT, DeLeon VB,Richtsmeier JT. Precision and error of three-dimensional pheno-typic measures acquired from 3dMD photogrammetric images.Am J Med Genet A 2005;138A:247-53.2. Kau CH, Richmond S, Incrapera A, English J, Xia JJ. Three-di-mensional surface acquisition systems for the study offacial mor-phology and their application to maxillofacial surgery. Int J MedRobotics Comput Assist Surg 2007;3:97-110.3. Ayoub A, Garrahy A, Hood C, White J, Bock M, Siebert JP, et al.Validation of a vision-based, three-dimensional facial imagingsystem. Cleft Palate Craniofac J 2003;40:523-9.4. Krimmel M, Kluba S, Bacher M, Dietz K, Reinert S. Digitalsurface photogrammetry for anthropometric analysis of the cleftinfant face. Cleft Palate Craniofac J 2006;43:350-5.5. Maal TJ, Plooij JM, Rangel FA, Mollemans W, Schutyser FA,Berge SJ. The accuracy of matching three-dimensional photo-graphs with skin surfaces derived from cone-beam computedtomography. Int J Oral Maxillofac Surg 2008;37:641-6.6. MoroA,CorreraP,BonielloR,GaspariniG,PeloS.Three-dimen-sional analysis in facial asymmetry: comparison with model anal-ysis and conventional two-dimensional analysis. J Craniofac Surg2009;20:417-22.7. Ferrario VF, Mian F, Peretta R, Rosati R, Sforza C. Three-dimen-sional computerized anthropometry of the nose: landmark repre-sentation compared to surface analysis. Cleft Palate Craniofac J2007;44:278-85.8. Weinberg SM, Scott NM, Neiswanger K, Brandon CA,Marazita ML. Digital three-dimensional photogrammetry: evalu-ation of anthropometric precision and accuracy using a Genex 3Dcamera system. Cleft Palate Craniofac J 2004;41:507-18.9. Khambay B, Nairn N, Bell A, Miller J, Bowman A, Ayoub AF.Validation and reproducibility of a high-resolution three-dimen-sional facial imaging system. Br J Oral Maxillofac Surg 2008;46:27-32.10. Rangel FA, Maal TJ, Berge´ SJ, van Vlijmen JC, Plooij JM,Schutyser F, et al. Integration of digital dental casts in 3-dimen-sional facial photographs. Am J Orthod Dentofacial Orthop2008;134:820-6.11. Bishara SE. Facial and dental changes in adolescents and theirclinical implications. Angle Orthod 2000;70:471-83.12. Winder RJ, Darvann TA, McKnight W, Magee JD, Ramsay-Baggs P. Technical validation of the Di3D stereophotogram-metry surface imaging system. Br J Oral Maxillofac Surg2008;46:33-7.13. WongJY,OhAK,OhtaE,HuntAT,RogersGF,MullikenJB,etal.Validity and reliability of craniofacial anthropometric measure-ment of 3D digital photogrammetric images. Cleft Palate Cranio-fac J 2008;45:232-9.14. Ferrario VF, Sforza C, Serrao G, Ciusa V, Dellavia C. Growth andaging of facial soft tissues: a computerized three-dimensionalmesh diagram analysis. Clin Anat 2003;16:420-33.15. FerrarioVF,SforzaC,PoggioCE,CovaM,TartagliaGM.Prelim-inary evaluation of an electromagnetic three-dimensional digi-tizer in facial anthropometry. Cleft Palate Craniofac J 1998;35:9-15.16. Sforza C, Peretta R, Grandi G, Ferronato G, Ferrario VF. Soft tis-sue facial planes and masticatory muscle function in skeletalClass III patients before and after orthognathic surgery treatment.J Oral Maxillofac Surg 2008;66:691-8.17. Dahlberg G. Statistical methods for medical and biologicalstudents. New York: Interscience Publications; 1940.18. Hammond P, Hutton TJ, Allanson JE, Campbell LE,Hennekam RC, Holden S, et al. 3D analysis offacial morphology.Am J Med Genet A 2004;126A:339-48.19. Celik E, Polat-Ozsoy O, Toygar Memikoglu TU. Comparison of cephalometric measurements with digital versus conventionalcephalometric analysis. Eur J Orthod 2009;31:241-6. 88 Rosati et al American Journal of Orthodontics and Dentofacial Orthopedics July 2010
