J Korean Academy of Advanced General Dentistry 2017;6:1-7 콘빔시티촬영프로토콜에따른 DICOM 파일의 STL 변환디지털모델의정확도비교 나지연 1, 차정열 2, 우창우 3, 김영현 1, 이정희 1, 황재준 1, 정호걸 1, 한상선 1 1 연세대학교치과대학영상치의학과, 2 연세대학교치과대학치과교정과, 3 연세대학교치과대학병원중앙기공실 ABSTRACT Comparative Accuracy of STL Conversion Digital Model of DICOM Files according to CBCT Scanning Protocols Ji Yeon Na 1, Jung-Yul Cha 2, Chang-Woo Woo 3, Young Hyun Kim 1, Jeong-Hee Lee 1, Jae Joon Hwang 1, Ho-Gul Jeong 1, Sang-Sun Han 1 1 Department of Oral and Maxillofacial Radiology, Yonsei University, College of Dentistry, Seoul, Korea 2 Department of Orthodontics, Yonsei University, College of Dentistry, Seoul, Korea 3 Central Dental Laoratory, Dental Hospital of Yonsei University, Seoul, Korea Ojectives: The purpose of this study is to evaluate the accuracy of the digital models from Cone eam computed tomography (CBCT) scans of orthodontic diagnostic models y comparing the Digital Imaging & Communication in Medicine (DICOM) file to Stereolithography (STL) conversion data otained y various voxel sized CBCT machines with the digital model otained y a 3D optical extraoral scanner. Materials and Methods: Three orthodontic diagnostic models were scanned with a 3D optical scanner and stored as STL file data. The three models were scanned with the 5 modes which were Oject scan (100 μm), 300, 400 μm of RAYSCAN α + and 300, 390 μm of Alphard 3030. The DICOM files were converted to STL file data. For each orthodontic model, two digital models otained y CBCT and optical scanner were paired, and GOM inspect software was used to superimpose and otain the deviation etween two digital models. One way ANOVA test was performed to determine whether there is a statistical difference in the mean value of deviations etween digital models y the 5 protocols. Results: In three diagnostic models, the mean values of the deviations etween the digital model of the Oject scan (100 μm) and the digital model of 3D optical scanner were the smallest in comparison with other image protocols and there was a statistically significant difference (p<0.001). The mean values of the deviations were in the range of 0.0967 ~0.1022 mm. On the other hand, deviations of STL data from 400 μm of RAYSCAN α + showed the largest mean value from 0.4007 to 0.4534 mm. Conclusions: The STL file data of the Oject scan (100 μm) of RAYSCAN α + is more accurate than those of large voxel sized protocols of CBCT. It shows the availaility as an orthodontic digital model. Key words : Digital model, CBCT scanning, STL conversion, Accuracy Correspondence : Prof. Sang-Sun Han Department of Oral and Maxillofacial Radiology, Yonsei University, College of Dentistry, 50-1 Yonsei-ro, Seodaemun-gu, Seoul, Korea Tel: +82-2-2228-3122, fax: +82-2-363-5232 E-mail: sshan@yuhs.ac Received: Feruary 24, 2017; Revised: April 18, 2017; Accepted: April 20, 2017 1
JOURNAL OF KOREAN ACADEMY OF ADVANCED GENERAL DENTISTRY 서론환자의구강내정보를재현하기위한치과모형의전통적인제작방식은인상채득후수작업으로석고모형을만드는것이다. 그러나석고모형은많은저장공간이필요하며파손및분실의가능성이크고, 변형시복구가불가능하다는단점이있다. 한편최근치과영역의디지털화가이루어지면서전통적인방법이아닌디지털모델에대한필요성이대두되었다 1. 구내스캐너를이용한구강내디지털인상채득, 구외스캐너와 Computer-aided design/computer-aided manufacture (CAD/CAM) 시스템을이용한보철물제작뿐아니라교정치료를위한계측치분석, 교정치료목적의 set-up과교정장치제작, 악교정수술에필요한수술용템플릿, computer guided 임플란트식립시뮬레이션과 implant guide surgery의임플란트수술용가이드제작에이르기까지최근치과치료전반에서디지털모델을이용한치료술식의사용분야가폭넓게확대되고있는추세이다 2-5. CAD/CAM에사용되는디지털데이터를획득하는방법에는직접법과간접법이있다 6. 직접법에는구내스캐너를이용하는방법이있으나스캐너의헤드가무겁고크기가커서구강이협소한환자나어린이에게서조작이어렵다는단점이있다. 숙련되지않은경우스캐너를적절하게위치시키기어려워영상이획득되지않는부분이생기기쉽고, 이를극복하기위해동일부위를반복스캔시영상의왜곡이발생할수있다 7. 또다른직접법에는 Cone Beam Computed Tomography (CBCT) 촬영데이터를 STL 형식으로변환하여디지털모델을제작하는방법이있다. 현재디지털모델을얻는가장보편적인방법은인상체에서얻은석고모형을 3D 광학스캐너로스캔하는것이다. 그러나광학스캐너는빛의조사각도나스캔되는피사체의 contour에영향을받기때문에주의를필요로하며 8,9 임상에서받아들여질만한정확도의디지털모델을얻으려면고가의광학스캐너의구비가필수적이다. 최근에는디지털모델을얻는간접법으로 CBCT 촬영영상의 STL 변환에대한연구가활발히진행되고있다 10,11. CBCT 스캔에기반을둔디지털모델은고가의광학스캐너의구비없이빠른시간내에데이터획득이가능하다는장점이있어교정용디지털모델제작에이용가능성이제시되고있다. Olivier de Waard 등은 Orthoproof사의 CBCT와 software을이용하여석고모형이나 인상체를스캔하고 CBCT data를 STL 파일로변환하여디지털모델을제작하는방법을소개한바있다 12. 그러나지금까지진행된 CBCT 스캔에관한연구는 1개의해상도를대상으로했다는한계점이있었다. 이에본연구에서는 CBCT 스캔의방법이해상도나장비에따라차이가있는지확인해보고자한다. 이연구의목적은 3개의교정진단용석고모형을이용하여 CBCT 스캔후얻은 DICOM 파일의 STL변환데이터와 3D 구외광학스캐닝으로얻은 STL 데이터의표면정확도차이를평가하고다양한촬영프로토콜로촬영한데이터의비교를통해프로토콜에따른 STL 변환값의정확도차이를검증해보고자한다. 재료및방법 1. 실험재료연세대학교치과대학병원치과교정과에 2015년 7월에서 9월사이교정치료를시작할목적으로내원한환자들의진단모형중지대치및결손치가없고총생이심하지않은 3개의성인하악교정진단용석고모형을난수표를이용하여무작위로선출하였다 (Fig 1). 2. 촬영프로토콜에따른 CBCT 스캐닝후 STL 변환 data 획득 CBCT 촬영장치 RAYSCAN α + (Ray Co., Ltd., Gyeonggi- Figure 1. The orthodontic diagnostic stone model. 2
Na et al. Accuracy of CBCT Scanning Digital Model Do, Korea) 의턱받이위에석고모형을고정한후 oject scan 모드를이용하여촬영하였다. 촬영조건은 90 kvp, 6 ma, voxel size 100 μm, field of view (FOV) 90 50 mm 2, 스캔시간은 14 sec으로숙련된촬영자에의해시행되었 다 (Fig 2). DICOM 파일형식으로획득된촬영데이터를 Raydent converter 소프트웨어 (Ray Co., Ltd., Gyeonggi-Do, Korea) 에전송하여 STL 파일형식으로변환하였다. 추가로 RAYSCAN α + 와 Alphard 3030 (Asahi Roentgen Figure 2. CBCT scanning. Figure 3. Scanning with optical extraoral scanner (Identica Hyrid, Medit, Seoul, Korea). [mm] 0.200 0.160 0.120 0.080 0.040 0.000-0.040-0.080-0.120 Z X Y - 0.160-0.200 Figure 4. Three-dimensional deviations etween test and reference STL file. Generated following prealignment y Gom inspect softwa re. Color-coded scale shows discrepancy of superimposition (mm). Tale 1. Detailed conditions according to CBCT scanning protocols Scanning protocol CBCT device Mode Voltage (kv) Current (ma) Exposure time (s) FOV (mm 2 ) Voxel size (μm) Oject scan (100 μm) RAYSCAN α + Oject scan 90 6 14 90 50 100 RAYSCAN α + 300 μm RAYSCAN α + Jaw Fast 90 4 4.9 80 100 300 RAYSCAN α + 400 μm RAYSCAN α + Sinus 90 6 14 140 100 400 Alphard 3030 300 μm Alphard 3030 P 80 8 17 154 154 300 Alphard 3030 400 μm Alphard 3030 C 80 10 17 200 200 390 3
JOURNAL OF KOREAN ACADEMY OF ADVANCED GENERAL DENTISTRY Ind., Co., Ltd., Kyoto, Japan) 의 voxel size 300 μm, 400 μm 의조건으로 3개의석고모형을각각촬영하였다. DICOM 파일로획득된촬영데이터는 Mimics 소프트웨어 (version 17.0; Materialise, Leuven, Belgium) 를이용하여 STL 파일형식으로변환되었다. 본연구에서사용된촬영프로토콜에따른세부촬영조건은 Tale 1과같다. 3. 3D 구외광학스캐닝 STL data 획득 CBCT로석고모형을촬영한같은날짜에각각의석고모형을 3D 광학구외스캐너 (Identica Hyrid, Medit, Seoul, Korea) 를이용하여디지털모델을얻었다 (Fig 3). 제조사가제공하는지침에따라스캔하였고얻어진데 이터는 STL 파일형식으로저장되었다. 4. 데이터분석분석소프트웨어는 GOM inspect (GOM GmH, Braunschweig, Germany) 를이용하였다. 각석고모형에대하여 CBCT와구외광학스캐너를이용하여만든디지털모델을한쌍으로하여구외광학스캐너에서얻은디지털모델을기준으로자동화된 prealignment의방식으로중첩하였다. 그리고두개의디지털모델간의차이값을각각의촬영프로토콜에서분석하였다. 석고모형별로양쪽제1대구치의최대풍융부를지나면서교합평면에평행한가상의단면 (Fig 5) 에서 5.8 mm간격으로분석점을설정하고, 구외광학스캐닝모델을기준으로각분석점에서의차이값을표면정확도로정의하였다 (Fig 6). 5. 통계분석 Figure 5. Reference plane for analysis. CBCT 스캐닝후 STL 변환 data와구외광학스캐닝 STL data로얻은디지털모델간의차이값을구하였고차이값의평균, 표준편차, 최소값, 최대값등의기술적통계량을제시하였다. 또한서로다른촬영프로토콜을통해얻은디지털모델의차이값의평균에통계학적인차이가있는지확인하기위하여 One way ANOVA test를실시하였다. 모든통계처리와분석은 SPSS (version 23.0; IBM, Armonk, NY) 통계처리프로그램을사용하여시행하였으며, 통계적유의성판정을위한유의수준은 0.05으로설정하였다. [mm] 0.200 0.160 0.120 0.080 0.040 0.000-0.040-0.080-0.120 Y Z X - 0.160-0.200 Figure 6. The deviation at each point is evaluated as the surface accuracy ased on the optical scanning model. 4
Na et al. Accuracy of CBCT Scanning Digital Model Tale 2. Comparison of deviations in the digital models according to CBCT scanning protocols Scanning protocol (Unit : mm) Model A Model B Model C Mean±SD Min Max p-value Mean±SD Min Max p-value Mean±SD Min Max p-value Oject scan (100 μm) 0.0967± 0.1022± 0.1017± a 0.001 0.363 a 0.004 0.310 a 0.0740 0.0113 0.0850 0.004 0.311 RAYSCAN α + 300 μm 0.3299± 0.1897 0.047 1.000 0.3307± 0.0157 c 0.115 0.581 0.3587± 0.1379 0.043 0.830 RAYSCAN α + 400 μm 0.4007± 0.2570 0.025 1.238 0.4087± <0.001 0.0231 c 0.060 0.699 0.4534± <0.001 0.1898 0.136 1.192 Alphard 3030 300 μm 0.2871± 0.1862 0.036 1.101 0.2748± 0.0188 0.071 0.522 0.3423± 0.1745 0.128 0.923 Alphard 3030 400 μm 0.3696± 0.4336 0.006 2.389 0.3756± 0.0388 c 0.001 1.280 0.3556± 0.2093 0.016 1.033 Values presented mean±sd p-value y one-way ANOVA, ac denoted y Bonferroni s test post-hoc analysis <0.001 결 과 고 찰 소프트웨어를이용하여 3개의석고모형의구외광학스캐너를이용하여얻은 STL 데이터를기준으로 CBCT 스캔의 STL 데이터를한쌍으로중첩하고분석한차이값을색으로나타냈다 (Fig 4). 중첩후디지털모델의양쪽제1대구치의최대풍융부를지나면서교합평면에평행한단면을차이값을분석하는기준단면으로설정하였다 (Fig 5). 이기준단면에서 5.8 mm의일정한간격으로분석점을설정하였고, 5가지의촬영프로토콜중어느하나에서라도차이값이측정되지않는점은분석에서제외하였다 (Fig 6). 모든분석점에서차이값은절대값으로환산되었다. 각모델에서얻어진차이값의평균, 표준편자, 최소값, 최대값은 Tale 2와같다. 측정된최대값은 0.310~2.389 mm 사이에존재하였고, 0.001~0.136 mm 사이의최소값을보였다. 또한평균값은 0.0967~0.4534 mm의범위로나타났다. Oject scan (100 μm) 의디지털모델과구외광학스캐닝디지털모델사이의차이값의평균은다른프로토콜에서얻은차이값과비교하여 0.0967~0.1022 mm 로가장작았으며 ANOVA 분석결과 3개의교정용석고모형모두에서통계학적으로유의한차이를나타냈다 (p<0.001). 반면 RAYSCAN α + 400 μm 프로토콜의차이값의평균은 0.4007~0.4534 mm로가장크게나타났다. 400 μm와 300 μm 각각에서 Alphard 3030의촬영장치를이용하여촬영한경우의평균이 RAYSCAN α + 촬영장치에서보다작게나타났다. 그러나같은장비로촬영한결과에서 400 μm에서 300 μm로 voxle size가작아질수록 3 개의석고모형모두에서평균과표준편차가감소하는경향을보였다. 3D 디지털모델은저장공간과파손및분실의가능성을줄일수있다는점외에도필요한모형을빠르고쉽게찾을수있고, 임상에서바로 3차원적접근이가능하다는장점이있다. 또한가상적모형을만들면비교및설명이쉽고, 협진시모형을전달할별도의운송수단없이전자우편등을이용한전송이가능하다 13,14. 그러나이러한장점에도불구하고디지털모델의해상도나재현성이떨어진다면보편적으로사용되기어려울것이다. 따라서일반진료실에서환자의구강을재현성있게구현하는디지털모델을얻기위해서는해상도높은고성능의광학스캐너를구비해야하며이를위해서는추가적으로높은비용이요구되는실정이다. 최근치과내임플란트치료가보편화됨에따라일반진료실의 CBCT 활용이증대되었다 15. 이러한시점에서 CBCT 스캔으로얻은디지털모델을 3D 광학스캐너로부터얻은디지털모델을기준으로비교하는것은고가의스캐너장비없이도임상적으로활용가능한디지털모델의제작가능성을확인해볼수있다는의의를갖는다. 그러나 Frits A. Rangel 등은 CBCT 스캔데이터와석고모형의스캔데이터간의평균차이가 0.6 mm였다고보고했다 16. 또한 Kang 등은 CT를통해얻은디지털모델은 voxel의크기, slice 두께에따른오차발생으로부정확할수있다고했다 10,17. 이와반대로 Wesemann 등은교정용모형을 CBCT 스캔하는것이임상적으로받아들여질만한정확도를갖는다고보고하였다 11. 이러한상반된연구는서로다른해상도의 CBCT 스캔데이터를이용하여연구함으로써발생한것으로생각된다. 아직까지디지털모델을위한 CBCT 스캔의해상도에따른정확도의차이 5
JOURNAL OF KOREAN ACADEMY OF ADVANCED GENERAL DENTISTRY 에대한연구는발표된바가없다. 본연구에서는서로다른해상도및장비로같은석고모형을 CBCT 스캔하여비교해본결과 CBCT의 voxel size가작을수록변환된디지털모델의표면정확도가개선되었다. 특히 100 μm로가장해상도가높았던 oject scan으로얻은디지털모델에서평가된표면정확도가다른 4가지의프로토콜보다우수하였으며통계학적으로유의한차이를보였다. 이는개선된해상도외에도 DI- COM 파일을 STL 파일로변환시표면데이터에서노이즈를제거하고, 인공음영에해당하는불분명한작은입자들을제거하는알고리즘이적용된 converter를이용하였기때문으로생각된다. Kazuo Hayashi 등은교정용진단모델에서임상적목적에부합하려면정확도가 0.1 mm 이내이어야한다고보고한바있다 18. 본연구에서 oject scan으로촬영하여얻은차이값의평균은 0.0967~0.1022 mm로가장작았으며 0.1 mm에근접하였다. 이러한결과는교정용디지털모델로써의활용가능성을보여준다고생각된다. 결론 CBCT의해상도가높을수록 STL로변환된디지털모델의표면정확도가개선되었다. CBCT의해상도와 STL 변환소프트웨어의개발이이루어질수록표면정확도는더욱개선될것으로예상된다. Acknowledgement The material is ased upon work supported the Ministry of Trade, Industry & Energy (MOTIE, Korea) under Advanced Technology Center Program. No.10062362, The devlopment of dental and medical prosthetics modeling, rapid farication and integrated trading system ased and converged on CBCT image, using Cloud networking 참고문헌 1. De Luca Canto G, Pachêco-Pereira C, Lagravere M, Flores-Mir C, Major P. Intra-arch dimensional measurement validity of laser- scanned digital dental models compared with the original plaster models: a systematic review. Orthod Craniofac Res 2015; 18(2):65-76. 2. Swennen GR, Mollemans W, De Clercq C, Aeloos J, Lamoral P, Lippens F, et al. A cone-eam computed tomography triple scan procedure to otain a three-dimensional augmented virtual skull model appropriate for orthognathic surgery planning. J Craniofac Surg 2009;20(2):297-307. 3. Sousa MVS, Vasconcelos EC, Janson G, Gari D, Pinzan A. Accuracy and reproduciility of 3-dimensional digital model measurements. Am J Orthod Dentofacial Orthop 2012;142(2):269-73. 4. Hassan WNW, Othman SA, Chan CS, Ahmad R, Ali SNA, Rohim AA. Assessing agreement in measurements of orthodontic study models: Digital caliper on plaster models vs 3-dimensional software on models scanned y structured-light scanner. Am J Orthod Dentofacial Orthop 2016;150(5):886-95. 5. 김종은, 김남훈, 박지현, 심준성. 임플란트가이드수술을위한 Planning 방식에대한고찰. 대한치과의사협회지 2016;54(2):108-22. 6. Kim J-H, Kim K-B. Evaluation of dimensional staility of digital dental model faricated y impression scanning method. J Dent Hyg Sci 2014;14(1):15-21. 7. 김려운, 장근원, 허유리, 손미경. 치과디지털인상의이해와적용. 대한치과재료학회지 2014;41(4):253-61. 8. Vogel AB, Kilic F, Schmidt F, Rüel S, Lapatki BG. Optical 3D scans for orthodontic diagnostics performed on full-arch impressions. J Orofac Orthop 2015;76(6):493-507. 9. Anh J-w, Park J-M, Chun Y-S, Kim M, Kim M. A comparison of the precision of three-dimensional images acquired y 2 digital intraoral scanners: effects of tooth irregularity and scanning direction. The Korean J Orthod 2016;46(1):3-12. 10. Kang S-H, Lee J-W, Lim S-H, Kim Y-H, Kim M-K. Dental image replacement on cone eam computed tomography with threedimensional optical scanning of a dental cast, occlusal ite, or ite tray impression. Int J Oral Maxillofac Surg 2014;43(10):1293-301. 11. Wesemann C, Muallah J, Mah J, Bumann A. Accuracy and efficiency of full-arch digitalization and 3D printing: A comparison etween desktop model scanners, an intraoral scanner, a CBCT model scan, and stereolithographic 3D printing. Quintessence Int 2016. 12. de Waard O, Rangel FA, Fudalej PS, Bronkhorst EM, Kuijpers- Jagtman AM, Breuning KH. Reproduciility and accuracy of linear measurements on dental models derived from cone-eam computed tomography compared with digital dental casts. Am J Orthod Dentofacial Orthop 2014;146(3):328-36. 13. DeLong R, Heinzen M, Hodges J, Ko C-C, Douglas W. Accuracy of a system for creating 3D computer models of dental arches. J Dent Res 2003;82(6):438-42. 6
Na et al. Accuracy of CBCT Scanning Digital Model 14. Lee G-T, Kim J-H, Kim W-C, Kim J-H. Three-dimensional evaluation on the repeataility and reproduciility of dental scannerased digital models. Journal of Korean Acedemy of Dental Technology 2012;34(3):213-20. 15. Zhao S, Roeltson D, Wang G, Whiting B, Bae KT. X-ray CT metal artifact reduction using wavelets: an application for imaging total hip prostheses. IEEE Trans Med Imaging 2000;19(12):1238-47. 16. Rangel FA, Maal TJJ, Bronkhorst EM, Breuning KH, Schols JGJH, Berge SJ, et al. Accuracy and Reliaility of a Novel Method for Fusion of Digital Dental Casts and Cone Beam Computed Tomography Scans. PLoS One 2013;8(3):e59130. 17. Widmann G, Stoffner R, Bale R. Errors and error management in image-guided craniomaxillofacial surgery. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;107(5):701-15. 18. Hayashi K, Sachdeva AU, Saitoh S, Lee S-P, Kuota T, Mizoguchi I. Assessment of the accuracy and reliaility of new 3-dimensional scanning devices. Am J Orthod Dentofacial Orthop 2013; 144(4):619-25. 7