Materials and methods

(1)

Claudia Spanier, Anja Ratzmann, Karl-Friedrich Krey

ΖQȵXHQFHRISULQWOD\HUKHLJKWDQGSULQWLQJ material on model accuracy and precision:

A 3D surface comparison of models SULQWHGXVLQJIXVHGȴODPHQWIDEULFDWLRQ

Claudia Spanier

Claudia Spanier, DMD

Orthodontist, Department of Orthodontics and Craniofacial Orthopaedics, University Medicine Greifswald, Greifswald, Germany

Anja Ratzmann, MSc

Senior Lecturer, Department of Orthodontics and Craniofacial Orthopae- dics, University Medicine Greifswald, Greifswald, Germany; Orthodontic practice, Wedel, Germany

Karl-Friedrich Krey, MME

Department Chair, Department of Orthodontics and Craniofacial Orthopaedics, University Medicine Greifswald, Greifswald, Germany;

Orthodontic practice, Wedel, Germany

Correspondence to: Prof Dr Karl-Friedrich Krey, Fleischmannstr. 42-44, 17475 Greifswald, Germany. Email: kreyk@uni-greifswald.de

KEY WORDS 'SULQWLQJ'VXSHULPSRVLWLRQGLJLWDOOLJKWSURFHVVLQJIXVHGȴODPHQWIDEULFDWLRQ orthodontic models

Objectives: 7R LQYHVWLJDWH WKH H΍HFW RI OD\HU KHLJKW RQ WKH DFFXUDF\RIRUWKRGRQWLFPRGHOVXWLOLVLQJIXVHGȴODPHQWIDEUL- FDWLRQSDUWLFXODUO\ZLWKUHJDUGWRRSWLPLVLQJLQRɝFHDOLJQHU PDQXIDFWXUH7KHVXLWDELOLW\RIIXVHGȴODPHQWIDEULFDWLRQZDV DVVHVVHGE\FRPSDULQJWKHUHVXOWVWRDKLJKSUHFLVLRQGLJLWDO light processing control group.

Materials and methods: Based on a digital sectioned maxil- ODU\PRGHOSK\VLFDOPRGHOVZHUHSULQWHGXVLQJIXVHGȴOD- PHQW IDEULFDWLRQ WHFKQRORJ\ DW GL΍HUHQW OD\HU KHLJKWV bwPwPwPwPwPwP wPwPDQGwPXVLQJWZRGL΍HUHQWPDWHU- LDOVSRO\ODFWLGH3/$1;DQGOLJQLQEDVHGSRO\PHU*UHHQ7(&

352 >([WUXGU /DXWHUDFK $XVWULD@ 7ZR '/3 PRGHOV ZLWK D OD\HUKHLJKWRIwPZHUHSURGXFHGUHSUHVHQWLQJWKHFRQ- WURO JURXS 6XEVHTXHQWO\ DOO SK\VLFDO PRGHOV ZHUH GLJLWDOO\

VFDQQHG DQG FRPSDUHG YLD ' VXSHULPSRVLWLRQ XVLQJ *20 ΖQVSHFWVRIWZDUH*20%UDXQVFKZHLJ*HUPDQ\

Results: 7KH 'DKOEHUJ DQDO\VLV DQG LQWUDREVHUYHU LQWUDFODVV FRUUHODWLRQSURYHGWKHDFFXUDF\RIWKH'VXSHULPSRVLWLRQPHDV- XUHPHQW WR EH H[FHOOHQW DQG UHSHDWDEOH 0RGHOV SULQWHG XVLQJ IXVHGȴODPHQWIDEULFDWLRQWHFKQRORJ\IURPOLJQLQEDVHGSRO\PHU ZLWKLQWKHUDQJHRIWRwPGHFUHDVHGLQSUHFLVLRQDV OD\HUKHLJKWLQFUHDVHG)XUWKHUPRUHWKHDQDO\VLVUHFRUGHGGH- FOLQLQJ SUHFLVLRQ RI IXVHG ȴODPHQW IDEULFDWLRQ PRGHOV EHORZ wP 0RGHOV SULQWHG XVLQJ OLJQLQEDVHG SRO\PHU ZHUH VXSHULRULQSUHFLVLRQFRPSDUHGWRWKRVHPDGHIURPSRO\ODFWLGH Conclusions: 7KH DFFXUDF\ DQG SUHFLVLRQ RI IXVHG ȴODPHQW IDEULFDWLRQPRGHOVFDQEHUHJXODWHGE\DOWHULQJOD\HUKHLJKW KRZHYHURWKHUSDUDPHWHUVVXFKDVRSWLPLVHGSULQWLQJPDWHU- LDODQGSULQWVHWWLQJVDUHQHFHVVDU\IRUFRQVLVWHQWKLJKTXDOLW\

$V VXFK IXVHG ȴODPHQW IDEULFDWLRQ SULQWLQJ LV DQ DFFXUDWH FRVWH΍HFWLYH DQG VXVWDLQDEOH WHFKQRORJ\ WR FUHDWH DOLJQHU models in orthodontic practice.

Introduction

As a result of the rapid technological advances that have WDNHQSODFHRYHUUHFHQWGHFDGHV'SULQWLQJLVQRZDYLDEOH RSWLRQLQRUWKRGRQWLFSUDFWLFH7KHV\PELRVLVRILQWUDRUDO scanning, virtual planning and appliance manufacturing R΍HUHG E\ WKLV WHFKQRORJ\ DOORZV IRU D FRPSOHWH GLJLWDO LQRɝFHZRUNȵRZ

(2)

7KHRULJLQVRIUDSLGSURWRW\SLQJGDWHEDFNWRZKHQ WKH-DSDQHVHDXWRPRELOHGHVLJQHU+LGHR.RGDPDLQYHQWHG an additive technology using ultraviolet light to cure poly- PHUVOD\HUE\OD\HUΖQ&KDUOHV+XOOHVWDEOLVKHGWKH ȴUVW'SULQWHUXWLOLVLQJVWHUHROLWKRJUDSK\6/$7KLVZDV IROORZHG E\ WKH GHYHORSPHQW RI GLJLWDO OLJKW SURFHVVLQJ '/3E\/DUU\+RUQEHFNLQIXVHGȴODPHQWIDEULFDWLRQ ))) E\ 6FRWW &UXPS LQ DQG WKH FRQFHSW RI LQN- MHWEDVHG'SULQWLQJDOVRNQRZQDV3RO\-HWSKRWRSRO\PHU printing (PPP), in 1998¹.

SLA, DLP, PPP and FFF play a key role in the creation of RUWKRGRQWLFGHQWDOPRGHOV7KH\PDLQO\GL΍HULQWHUPVRI

print resolution, printing speed, and the cost of the technology itself and its associated materials. Other factors include SULQW YROXPH SULQWLQJ RULHQWDWLRQ FDUERQ IRRWSULQW DQG SRVWSURFHVVLQJSURFHGXUHV3ULQWUHVROXWLRQZKLFKFDQEH DGMXVWHG E\ DOWHULQJ WKH OD\HU KHLJKW KDV EHHQ IRXQG WR have a particular impact on the accuracy of dental casts². Previous studies found a higher Z-resolution, which equates WRDUHGXFHGOD\HUKHLJKWWREHFRUUHODWHGZLWKKLJKHUDF- FXUDF\ RI WKH SULQWHG REMHFW^2,3. Interestingly, decreasing OD\HU KHLJKW OHDGV WR D KLJKHU DPRXQW RI PDWHULDO WR EH printed and exponentially higher printing times (Fig 1, 7DEOH UHVXOWLQJ LQ KLJKHU RYHUDOO PRGHOOLQJ FRVWV⁴.

)LJbbPrinting time according to layer height.

50.00 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00

Printing time (hours)

Layer height (μm)

0.0 50.0 100.0 150.0 200.0 250.0 300.0

For one model For three models For six models For nine models

7DEOHbbSimulation of printing times in relation to Z-resolution for the TEVO Tornado FFF printer

Layer height (μm) Printing time for one model Printing time for nine models

50.0 4 h 55 min 44 h 58 min

100.0 2 h 27 min 22 h 28 min

150.0 1 h 38 min 14 h 58 min

200.0 1 h 14 min 11 h 14 min

250.0 1 h 0 min 9 h 10 min

300.0 49 min 7 h 31 min

(3)

Consequently, FFF printing with as low a Z-resolution as is FOLQLFDOO\SRVVLEOHLVRIFUXFLDOLPSRUWDQFHWRHQDEOHFRVW HɝFLHQWLQRɝFHDOLJQHUSURGXFWLRQ

Taking into account the economic advantages and sim- plicity of use of FFF printers, it is surprising that numerous studies have examined the accuracy of dental models printed using SLA, DLP and PPP technology^3,5-9, whereas there is little research on FFF technology^2,4,10. Concerning FFF printing, Kamio et al⁴ XWLOLVHG ZKROH PDQGLEOHV ZLWK layer heights from 200 to 500 μm, Lee et al¹⁰ used single replica teeth with a layer height of 330 μm, and Pérez et al² focused on various printing parameters, working with cylin- drical samples and layer heights of 150 and 250 μm.

The aims of the present study were twofold. First, the H΍HFWRI=UHVROXWLRQRQWKHDFFXUDF\RIRUWKRGRQWLFPRGHOV printed using FFF technology was examined utilising a sectioned maxillary model with layer heights ranging from 50.0 WRwP6HFRQGWKHFOLQLFDOVXLWDELOLW\RI)))SULQWLQJ ZDVHYDOXDWHGE\FRPSDULQJWKHLUDFFXUDF\WRDKLJKSUHFL- sion DLP control group with a layer height of 20 μm.

Materials and methods

To examine the quality of the models printed using FFF, a maxillary arch was taken from a randomly selected digital GHQWDOPRGHODQGPRGLȴHGLQ2Q\[&HSK'/DEΖPDJHΖQ- VWUXPHQWV&KHPQLW]*HUPDQ\E\VOLFLQJDWWKHERWWRPRI WKHJLQJLYDDQGGLVWDOO\IURPWKHPD[LOODU\ULJKWȴUVWSUHPRODU DQGPD[LOODU\OHIWFHQWUDOLQFLVRU6XEVHTXHQWO\DGGLWLYHDW- WDFKPHQWVDQGDVXEWUDFWLYHUHFHVVZHUHDGGHGWRWKLVVHF- tioned digital model. With the aid of the resulting master STL ȴOH)LJWZRLGHQWLFDOSK\VLFDOPRGHOVZLWKDOD\HUKHLJKW of 20.0 μm were printed using DLP technology (SprintRay, /RV$QJHOHV&$86$ZLWKGLHDQGPRGHOUHVLQSURYLGHGE\

the same company) (Fig 2), representing the control group.

Then, 18 sectioned maxillary models were produced with FFF printing (TEVO Tornado, TEVO 3D Electronic Technol ogy,

=KDQMLDQJ&KLQDZLWKWZRGL΍HUHQWELRSRO\PHUVWKHSRO\- ODFWLGH 3/$ 1; DQG WKH OLJQLQEDVHG SRO\PHU *UHHQ7(&

PRO (Extrudr, Lauterach, Austria) (Fig 2), each divided into QLQHGL΍HUHQWJURXSVwPwPwPwP wPwPwPwPDQGwP

All the physical maxillary models were then digitised XVLQJD'PRGHOVFDQQHU6$UWL=LUNRQ]DKQ*DLVΖWDO\

resolution 10 μm) to produce stereolithography (STL) test ȴOHV8WLOLVLQJ*20ΖQVSHFW*20%UDXQVFKZHLJ*HU- PDQ\WKHWHVWȴOHVZHUHVXSHULPSRVHGRQWRWKH67/PDVWHU ȴOHZLWKWKHDLGRIDQDXWRPDWHGEHVWȴWDOJRULWKPPDWFKLQJ the two virtual models according to the characteristics of the teeth. Applying the module “Surface comparison to CAD”, the accuracy was evaluated using measurement tools analysing SRLQWGHYLDWLRQVDQGDOVRYLVXDOO\XVLQJDFRQWLQX- RXVFRORXUVSHFWUXP%OXHLVKQXDQFHVUHYHDOHGGHȴFLHQFLHV of the scanned model surface in comparison to the master ȴOHZKHUHDVUHGGLVKQXDQFHVLQGLFDWHGDQH[FHVVRIVFDQQHG material and green indicated measurement agreement.

With reference to previous studies^3,8,11,12, the critical threshold was set at 0.25 mm. Using the inspection tool, arithmetic PHDQ$0VWDQGDUGGHYLDWLRQ6'PLQLPXPDEVROXWHGH- YLDWLRQDQGPD[LPXPDEVROXWHGHYLDWLRQZHUHFDOFXODWHG 7KHVHYDOXHVZHUHJDLQHGE\PHDVXULQJWKHRUWKRJRQDOGLV- WDQFHEHWZHHQWKHFRUUHVSRQGLQJSRLQWVRIWKH&$'SRO\JRQ PHVKDQGWKHSRLQWFORXGRIWKHWHVWȴOH6XEVHTXHQWO\UH- ports were drawn up from each 3D superimpos ition, includ- ing colour maps and measurement data (Figs 3 and 4). In the LQWHUHVWRIH[DPLQLQJWKHUHOLDELOLW\RIWKH'VXSHULPSRV LWLRQ PHWKRG RI PHDVXUHPHQW DOO WKH WHVW ȴOHV WKDW RULJL- QDWHGIURPWKHPRGHOVSULQWHGXVLQJ)))DQGOLJQLQEDVHG polymer were measured twice.

Statistical analysis

To evaluate the trueness of the dental models produced, the AMs of the deviation of the corresponding points of the VXSHULPSRVHGVXUIDFHVRIWKHWHVWDQGPDVWHUȴOHVZHUH DQDO\VHG3UHFLVLRQZDVHVWLPDWHGE\DVVHVVLQJWKH6'RI WKH GLVFUHSDQF\ EHWZHHQ WKH FRPSDUHG VXUIDFHV RI WKH ȴOHV)RUIXUWKHUHYDOXDWLRQWKHSHUFHQWDJHRISRLQWVZLWKLQ WKH FULWLFDO ERXQGV RIsPP DQG ZLWKLQ WKH QRPLQDO ERXQGVRIsPPZHUHDQDO\VHGEDVHGRQWKHQRUPDO- ity of measurement points¹³. With the aid of the colour map analysis of the 3D superimposition, information was gained concerning the location and degree of deviation or congru- HQFHRIWKHFRUUHVSRQGLQJVXUIDFHV5HOLDELOLW\ZDVHYDOX- DWHGXVLQJ63666WDWLVWLFVYHUVLRQΖ%0$UPRQN 1<86$)LUVWWKHLQWUDFODVVFRUUHODWLRQFRHɝFLHQWΖ&&RI UHSHDWHGPHDVXUHPHQWVIRUDVLQJOHREVHUYHURQWKHEDVLV RIDEVROXWHDJUHHPHQWZDVFDOFXODWHG6HFRQGWKH'DKO- EHUJHUURUZDVDQDO\VHGWRDVVHVVYDULDELOLW\GXHWRWHFK nical inconsistencies.

(4)

a

j d

m g

p

b

k e

n h

q

c

l f

o i

r

(5)

)LJVDWR[bbSectioned maxillary dental model: (a to c)67/PDVWHUȴOH(d to f) DLP control model (layer height 20.0 μm);

(g to i)/LJQLQEDVHGPRGHOOD\HUKHLJKWwP(j to l) Polylactide model (layer height 50.0 μm); (m to o)/LJQLQEDVHGPRGHOOD\HU height 150.0 μm); (p to r) Polylactide model (layer height 150.0 μm); (s to u)/LJQLQEDVHGPRGHOOD\HUKHLJKWwP

(v to x) Polylactide model (layer height 300.0 μm).

)LJVDWRFbb'VXSHULPSRVLWLRQFRORXUPDSDQDO\VLVRIWHVWȴOHVDQG&$'UHIHUHQFHȴOH(a) DLP control model (layer height 20.0 μm);

(b)/LJQLQEDVHGPRGHOOD\HUKHLJKWwP(c) Polylactide model (layer height 100.0 μm).

s

v

t

w

u

x

a b c

ȫ ȫ ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

(6)

)LJVDWRKbb'VXSHULPSRVLWLRQFRORXUPDSDQDO\VLVRI)))SULQWHGOLJQLQEDVHGGHQWDOPRGHOZLWKGL΍HUHQWOD\HUKHLJKWVDQG&$' UHIHUHQFHȴOH(a and b) Layer height 100.0 μm; (c and d) Layer height 150.0 μm; (e and f) Layer height 200.0 μm; (g and h) Layer height 250.0 μm.

a

c

e

g

b

d

f

h

ȫ

ȫ ȫ

ȫ

ȫ ȫ ȫ

ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫȫ

ȫ ȫ

ȫ

ȫ ȫ

ȫ

(7)

Results

7KHUHOLDELOLW\H[DPLQDWLRQRIWKH'DQDO\VLVLVVKRZQLQ 7DEOH )URP WKH YDOXHV RI WKH LQWUDREVHUYHU Ζ&&V Ζ&&

$0bΖ&&6'LWFDQEHVWDWHGWKDWWKHDSSOLHGPHDV- XUHPHQWPHWKRGYLD'VXSHULPSRVLWLRQKDVKLJKUHOLDELOLW\

0RUHRYHU KDYLQJ TXDQWLȴHG WKH WHFKQLFDO PHDVXUHPHQW HUURU E\ LPSOHPHQWLQJ WKH 'DKOEHUJ IRUPXOD 'DKOEHUJ HUURU$0ȯPP'DKOEHUJHUURU6'ȯPPWKH H[FHOOHQW VXLWDELOLW\ RI ' DQDO\VLV XVLQJ *20 ΖQVSHFW LV reinforced.

The outcome of the comparison of the 3D superimpos- LWLRQRIWHVWDQGVRXUFHȴOHVLVVXPPDULVHGLQ7DEOH)XU- ther statistical calculations of the percentage of points ZLWKLQQRPLQDOERXQGVIRUWKHOLJQLQEDVHGSRO\PHUPRGHOV printed using FFF and the DLP control group are presented LQ7DEOH

Examining the parameters of accuracy, namely the AM, 6'DQGSHUFHQWDJHRISRLQWVZLWKLQWKHFULWLFDOERXQGVWKH RYHUDOOGL΍HUHQFHVEHWZHHQWKHH[SHULPHQWDOJURXSV))) SULQWHGOLJQLQEDVHGSRO\PHU)))SULQWHGSRO\ODFWLGHDQG '/3FRQWUROJURXSZHUHGHWHUPLQHG7DEOH

The AM of the deviation of the corresponding points of WKHVXSHULPSRVHGVXUIDFHVUDQJHGIURPȫWRȫPP LQ WKH JURXSV WKDW XVHG ))) SULQWLQJ DQG IURP ȫ WR ȫPPLQWKH'/3FRQWUROJURXS&RQFHUQLQJWUXHQHVV FFF printed models seemed to have smaller overall dimensions¹⁰ZKHUHDVWKRVHIDEULFDWHGXVLQJ'/3SULQWLQJRQO\

had slightly smaller dimensions.

ΖQWHUPVRISUHFLVLRQWKHOLJQLQEDVHGSRO\PHUPRGHOV printed using FFF displayed overall lower SDs and a higher DPRXQWRIPHDVXUHPHQWSRLQWVZLWKLQWKHFULWLFDOERXQGV RIsPPWKDQWKHSRO\ODFWLGHPRGHOVSULQWHGXVLQJ))) 7DEOH)LJ:KHQFRPSDUHGWRWKH'/3FRQWUROJURXS WKHSUHFLVLRQUHTXLUHPHQWVZHUHRQO\PHWE\OLJQLQEDVHG PRGHOVZLWKOD\HUKHLJKWVEHWZHHQDQGwPFRQ- sidering the SD and percentage of points within the critical ERXQGV ! 0RUHRYHU DOO WKH OLJQLQEDVHG PRGHOV printed using FFF, with the exception of the model with a OD\HUKHLJKWRIwPKDGRYHURISRLQWVZLWKLQWKH FULWLFDOERXQGVGLVSOD\LQJDKLJKOHYHORIFRQVLVWHQF\RYHUD wide range of layer heights (50.0 to 300.0 μm). Interestingly, only the FFF printed polylactide model with a layer height of 250.0 μm also met these requirements.

7DEOHbb5HOLDELOLW\RI'VXSHULPSRVLWLRQPHWKRGRIPHDVXUHPHQWLQ*20ΖQVSHFWIRU$0DQG6' Layer height (μm) AM measurement 1

(mm) AM measurement 2

(mm) SD measurement 1

(mm) SD measurement 2

(mm)

50.0 ȫ ȫ 0.11 0.11

80.9 ȫ ȫ 0.10 0.10

100.0 ȫ ȫ 0.09 0.09

150.0 ȫ ȫ 0.10 0.10

ȫ ȫ 0.10 0.10

200.0 ȫ ȫ 0.12 0.11

250.0 ȫ ȫ 0.12 0.12

300.0 ȫ ȫ 0.12 0.12

ȫ ȫ 0.13 0.13

'DKOEHUJHUURUPP 0.002357 0.002357

Ζ&&DEVROXWHDJUHHPHQW 0.900

(8)

7DEOHbbComparison of models printed using FFF with lignin- EDVHGSRO\PHUDQGWKH'/3FRQWUROJURXSEDVHGRQWKH SHUFHQWDJHRISRLQWVZLWKLQWKHQRPLQDOERXQGV 7DEOHbbMeasurement data for the 3D superimposition and

SHUFHQWDJHRISRLQWVZLWKLQWKHFULWLFDOERXQGVDVDIXQFWLRQRI layer height, technology and material of the dental models studied

Material/

technol- ogy

Layer height (μm)

AM (mm) SD

(mm) Points within critical bounds

± 0.25 mm (%) Lignin-

EDVHG))) 50.0 ȫ 0.11 97.18

80.9 ȫ 0.10 98.02

100.0 ȫ 0.09 99.18

150.0 ȫ 0.10 98.02

ȫ 0.10 98.35

200.0 ȫ 0.12

250.0 ȫ 0.12

300.0 ȫ 0.12

ȫ 0.13 94.28

Poly- ODFWLGH FFF

50.0 ȫ 0.22 74.31

80.9 ȫ 0.19 80.91

100.0 ȫ 0.18 82.94

150.0 ȫ 0.20

ȫ 0.20

200.0 ȫ 0.21

250.0 ȫ 0.12

300.0 ȫ 0.24 70.02

ȫ 0.15 89.85

Control

JURXS'/3 20.0 ȫ 0.10 98.58

20.0 ȫ 0.10 98.71

)LJbb5HODWLRQEHWZHHQ layer height and SD with an LQFUHDVHEHORZwP DQGDERYHwP through the example of FFF SULQWHGOLJQLQEDVHG models.

0.14 0.13 0.12 0.11 0.10 0.09 0.08

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35

SD (mm)

Layer height (mm)

Material/

technology Layer height

(μm) Points within

nominal bounds

± 0.05 mm (%) Lignin-

EDVHG))) 50.0 33.87

80.9 35.57

100.0 40.04

150.0 35.57

200.0

250.0

300.0 31.77

Control

JURXS'/3 20.0 37.59

20.0 38.11

(9)

$IWHUH[DPLQLQJWKHLQȵXHQFHRI=UHVROXWLRQLQHDFKRI WKHH[SHULPHQWDOJURXSVVRPHDVVXPSWLRQVFDQEHPDGH 7DEOHΖQWKHOLJQLQEDVHGJURXSWKHPRVWDFFXUDWHDQG precise values were reached at a layer height of 100.0 μm

$0ȫPP6'PPRIGDWDSRLQWVZLWKLQ WKHFULWLFDOERXQGVDQGZLWKLQWKHQRPLQDOERXQGV even surpassing the precision parameters of the DLP con- WUROJURXS7DEOHΖQFRQWUDVWWKHORZHVWFRQVLVWHQF\ZDV IRXQG DW D OD\HU KHLJKW RI wP $0 ȫPP 6'b PP RI GDWD SRLQWV ZLWKLQ WKH FULWLFDO ERXQGVΖQWHUHVWLQJO\WKHEHVWUHVXOWVIRUFRQVLVWHQF\LQ WKH)))SULQWHGSRO\ODFWLGHJURXSZHUHREVHUYHGDWDOD\HU KHLJKW RI wP 6' PP RI GDWD SRLQWV ZLWKLQWKHFULWLFDOERXQGVZKHUHDVSRO\ODFWLGHPRGHOVZLWK D OD\HU KHLJKW RI wP $0 ȫPP 6' PP RIGDWDSRLQWVZLWKLQWKHFULWLFDOERXQGVZHUHWKH least accur ate in their experimental group.

Analysing the SD independently of the layer height of WKHOLJQLQEDVHGPRGHOVSULQWHGXVLQJ)))DQLQFUHDVHLQ6' ZDV REVHUYHG DV OD\HU KHLJKW LQFUHDVHG IURP WR bwP)LJVDQGZKHUHDVWKH6'GHFUHDVHGDVOD\HU height increased from 50.0 to 100.0 μm. Aside from the FRUUHODWLRQ EHWZHHQ OD\HU KHLJKW DQG 6' D GHSHQGHQFH ZDVDOVRREVHUYHGEHWZHHQWUXHQHVVDQGOD\HUKHLJKWLQWKH OLJQLQEDVHGJURXSUHSUHVHQWLQJDVOLJKWO\LQFUHDVLQJ$0 with increasing layer height. In the FFF printed polylactide JURXSDVLPLODUUHODWLRQZDVIRXQGEHWZHHQ6'DQGOD\HU height with the exception of layer heights of 250.0 and wP7DEOH

With the aid of the colour map analysis (Figs 3 and 4), the extent and location of the deviation of the corresponding VXUIDFHV RI WKH WHVW DQG VRXUFH ȴOH FRXOG EH H[SORUHG Greenish areas indicated an excellent match of the com- SDUHG VXUIDFHV ZLWKLQ WKH WROHUDWHG ERXQGV D WUDQVLWLRQ LQWREOXHQXDQFHVLQGLFDWHGGHȴFLHQFLHVRUVPDOOHUGLPHQ- VLRQVRIWKHWHVWHGVXUIDFHLQUHODWLRQWRWKHVRXUFHȴOHDQG reddish areas represented an excess of scanned material.

Generally, very precise greenish areas were found on cusp VORSHV DQG YHVWLEXODU DQG RUDO VPRRWK VXUIDFHV %OXHLVK colour patches were detected interdentally, at the cervix GHQWLVDQGLQFLVDOHGJHVDQGRQWKHYHVWLEXODURUDOPHVLDO and gingival attachment surfaces. Reddish nuances, namely excessive dimensions, were found on the occlusal and dis- WDODWWDFKPHQWDUHDVRFFOXVDOȴVVXUHVFXVSWLSVFDYLW\VXU- faces, and interdentally.

Discussion

7KHSUHVHQWVWXG\DVVHVVHGWKHLQȵXHQFHRIOD\HUKHLJKWRQ the accuracy of FFF printed dental models applying a 3D VXSHULPSRVLWLRQDQGLQYHVWLJDWHGWKHFOLQLFDOVXLWDELOLW\RI )))SULQWLQJE\FRPSDULQJWKHSULQWLQJTXDOLW\WR'/3WKH gold standard.

When assessing trueness and precision, the model in SULQWHGVFDQQHGDQG67/ȴOHIRUPZDVFRPSDUHGWRWKH VRXUFH ȴOH PHDVXULQJ SRLQW GHYLDWLRQV EHWZHHQ WKH WHVW DQG PDVWHU ȴOH LQ ERWK QHJDWLYH DQG SRVLWLYH GLUHFWLRQV 7DNLQJWKH'DKOEHUJHUURUDQGWKHLQWUDREVHUYHUΖ&&LQWR DFFRXQWDQH[FHOOHQWPHDVXUHPHQWPHWKRGFDQEHDVFHU- WDLQHG7DEOHKRZHYHUDGGLWLRQDOVRXUFHVRIHUURUZHUH encountered during the scanning process that were not inspected in the present study. First, since the model scan XWLOLVHGDOLJKWEHDPWKDWGLVSHUVHGOLQHDUO\FHUWDLQORFD- WLRQV ZHUH DW JUHDWHU ULVN RI VFDQQLQJ HUURU VXFK DV RE- scured surfaces, namely occlusal grooves, interdental spaces and retractions on attachments^10,14. Thus, to avoid DUWHIDFWV VFDQQLQJ LPDJHV WDNHQ IURP GL΍HUHQW DQJOHV ZHUH FRPELQHG 6HFRQG WKH WUDQVIRUPDWLRQ RI WKH VFDQ GDWDLQWRDQ67/ȴOHPD\KDYHFDXVHGHUURUVGXHWRGDWD conversion¹⁰ 1RQHWKHOHVV WKH FOLQLFDO VXLWDELOLW\ RI WKH 6$UWLPRGHOVFDQQHUZDVSURYHQLQDSUHYLRXVVWXG\¹⁵. Interestingly, the increase in accuracy that was antici- pated to occur with a decrease in layer height, i.e., an increase in Z-resolution, did not entirely occur with the FFF SULQWHGVHTXHQWLDOGHQWDOPRGHOV:LWKWKHOLJQLQEDVHG group in particular, a continuous improvement in accuracy with regard to SD and the percentage of points within FULWLFDOERXQGVZDVQRWHGDVOD\HUKHLJKWGHFUHDVHGZLWKLQ WKHUDQJHRIWRwP7DEOH:KHQOD\HUKHLJKW GHFUHDVHGEH\RQGwPKRZHYHUDFFXUDF\DOVRGH- FUHDVHG)LJΖQJHQHUDOWKHUHDSSHDUHGWREHDQRSWL- PDOOD\HUKHLJKWRIwPLQWKHOLJQLQEDVHGJURXS which was not found in the highest Z-resolution recom- mended in the manufacturer’s instructions for the FFF SULQWHU7KLVPD\KDYHEHHQEHFDXVHRQWKHRQHKDQG reducing the height of each layer leads to an increase in WKHQXPEHURIOD\HUVDQGKHLJKWHQVWKHULVNRISULQWLQJ errors such as artefacts or failure during the printing process itself⁸7RLOOXVWUDWHWKLVSRLQWDOD\HUKHLJKWRIbwP has six times more layers than a dental cast with a layer height of 300.0 μm, and the former increases the likeli-

(10)

hood of printing errors simply due to the additional num- EHU RI OD\HUV WR EH SULQWHG 2Q WKH RWKHU KDQG WKH ))) SULQWHU XVHG H[KLELWHG REYLRXV GLɝFXOWLHV LQ SXOOLQJ WKH previous layers from the printing platform due to an in- DFFXUDWHGLVWDQFHEHWZHHQWKHQR]]OHDQGWKHSODWIRUPDW WKH EHJLQQLQJ RI WKH SULQWLQJ SURFHVV ZKHQ SULQWLQJ smaller layer heights such as 50.0 and 80.9 μm. It was GLɝFXOWWROHYHOWKHSULQWEHGLQȴUVWOD\HUGLVWDQFHVXQGHU ƉPLQSUDFWLFDOKDQGOLQJHYHQLIWKHSULQWHUEHLQJ XVHGZDVHTXLSSHGZLWKDQDXWROHYHOOLQJV\VWHP%/7RXFK

$QWFODEV6HRXO6RXWK.RUHD

&RPSDULQJWKHDFFXUDF\EHWZHHQERWK)))SULQWHGH[- perimental groups with regard to printing material, models PDGHIURPOLJQLQEDVHGSRO\PHUKDGDFRQVLVWHQWO\ORZHU SD and thus more measurement points within the clinical ERXQGVWKDQWKH3/$PRGHOV7DEOH)LJDVVXFKWKH SULQWLQJPDWHULDODOVRVHHPHGWRD΍HFWDFFXUDF\$SUHYLRXV VWXG\ IRXQG WKDW ERWK SRO\ODFWLGH DQG OLJQLQEDVHG SRO\- mers have excellent printing properties 'L΍HUHQFHV could arise due to temperature resistance, as indicated on WKHGDWDVKHHWVIRUWKHPDWHULDOVSURYLGHGE\WKHPDQXIDF- turer^17,18 7KH OLJQLQEDVHG SRO\PHU *UHHQ7(& 352 UH- ceived a maximum of 10 points for temperature resistance according to the data sheet, whereas the polylactide PLA X2 only received 4 points^17,18/LNHZLVHWKHOLJQLQEDVHGSRO\

mer scored slightly higher in the categories of impact resistance and maximum stress than the polylactide did. Equal values were recorded for visual quality, layer adhesion and HORQJDWLRQDWEUHDNΖQJHQHUDOEHWWHUDFFXUDF\VHHPHGWR DULVH GXH WR WKH EHWWHU PDWHULDO DWWULEXWHV RI WKH OLJQLQ EDVHGSRO\PHUXWLOLVHG^17,18.

ΖQWHUPVRIFOLQLFDOVXLWDELOLW\LWZRXOGEHLQWHUHVWLQJWR NQRZKRZDFFXUDWHDQGSUHFLVHGHQWDOFDVWVQHHGWREHWR ensure the delivery of successful orthodontic therapy with aligners; however, there is currently no consensus concerning accuracy. Previous studies set limits of clinical agreement ranging from 0.2 to 0.5 mm^19-21. Given that a consid- HUDEOH QXPEHU RI SUHYLRXV VWXGLHV VHW WKHLU FOLQLFDO threshold at 0.25 mm3,5,8,10-12, the present study did the same. One reason for which a deviation of 0.25 mm was DFFHSWHG ZDV WKDW WKH $PHULFDQ %RDUG RI 2UWKRGRQWLFV

*UDGLQJ6\VWHP$%22*6HVWDEOLVKHGWRHYDOXDWHGHQWDO FDVWVIRUȴQLVKHGRUWKRGRQWLFWUHDWPHQWVFRQVLGHUVDGHYL- DWLRQRIXSWRPPWREHFOLQLFDOO\VXLWDEOHLQWHUPVRI alignment and marginal ridges^3,22. The 3D superimposition

algorithm applied compared the point deviations of corres- SRQGLQJ VXUIDFHV ZKHUHE\ D PD[LPXP GHYLDWLRQ RI PPLQERWKDSRVLWLYHDQGQHJDWLYHGLUHFWLRQZRXOG equal a linear deviation of 0.50 mm maximum according to WKH$%22*61RQHWKHOHVVIXUWKHUVWXGLHVDUHUHTXLUHGWR GHȴQH D UHDVRQDEOH ERXQGDU\ IRU FOLQLFDO VXLWDELOLW\ GH- pending on the actual incoming transmission of tooth movement from the printed dental model to the vacuum- formed aligner.

7KH DGYDQWDJHV RI ))) SULQWLQJ DUH WKH FRVWH΍HFWLYH DFTXLVLWLRQDQGPDLQWHQDQFHRIWKHSULQWHUKLJKYDULDELOLW\

and duration of the printing materials, ease of hand ling, WLPHH΍HFWLYHQHVVLQSURGXFWLRQDQGDGHTXDWHUHOLDELOLW\RI the printing results². Moreover, increased layer height of- IHUVVLJQLȴFDQWHFRQRPLFEHQHȴWVGXHWRWKHVOLJKWO\ORZHU ȴODPHQWFRQVXPSWLRQDQGH[SRQHQWLDOO\VKRUWHUSULQWLQJ times (Fig 1)⁴7KXVSULQWLQJWLPHGRXEOHVZKHQOD\HUKHLJKW decreases from 100.0 to 50.0 μm; as such, the total produc- WLRQWLPHIRUQLQHwPGHQWDOPRGHOVZRXOGEHKRXUV DQG PLQXWHV ZKHUHDV SULQWLQJ WKH VDPH QXPEHU RI PRGHOVZLWKDOD\HUKHLJKWRIbwPZRXOGWDNHKDOIWKH time, namely 22 hours and 28 minutes. For a layer height of 300.0 μm, printing nine models would take no longer than 7 hours and 31 minutes, which is six times less time than that required to print nine models with a layer height of wP7DEOH

Although printing dental models with a high Z-resolution such as 50.0 μm is a more time-consuming process, LWLVQRWQHFHVVDULO\MXVWLȴHGE\SURSRUWLRQDOO\KLJKHUDFFXU acy. Despite the fact that the most accurate and precise printing result in the present study was found in the lignin- EDVHGJURXSDWDOD\HUKHLJKWRIwP$0ȫPP 6' PP RI GDWD SRLQWV ZLWKLQ WKH FULWLFDO ERXQGVDQGZLWKLQWKHQRPLQDOERXQGVWKHEHQH- ȴWJDLQHGLQDFFXUDF\ZDVQRWLQUHDVRQDEOHSURSRUWLRQWR a printing time over 1.5 times longer compared to a layer KHLJKWRIwP$0PP6'PPRI GDWDSRLQWVZLWKLQWKHFULWLFDOERXQGVDQGZLWKLQWKH QRPLQDOERXQGV%DVHGRQWKLVLWZRXOGEHLQWHUHVWLQJWR determine whether even models with a layer height of wP $0 ȫPP 6' PP RI GDWD SRLQWV ZLWKLQ WKH FULWLFDO ERXQGV DQG ZLWKLQ WKH QRPLQDOERXQGVWUDQVIRUPDGHTXDWHIRUFHVWRWKHWRRWK using vacuum-formed aligners. Further research is required IRUFODULȴFDWLRQ

(11)

$FOLQLFDOVWXG\E\'DYLVDWDO²³ focused on the potential health concerns arising from volatile gases and particles during the FFF printing process. The commonly determined KD]DUGRXV YRODWLOH FRPSRXQGV HPLWWHG E\ ))) SULQWHUV were formaldehyde, a human carcinogen; styrene and PHWK\OHQHFKORULGHFRQVLGHUHGSUREDEO\FDUFLQRJHQLFIRU KXPDQVDQGWROXHQHDWR[LFK\GURFDUERQ²³. Nevertheless, the total volatile air compound emissions (TVOC ERs) were generally two orders of magnitude lower than those from dry process copiers, laser printers and personal com- puters²³$PRQJWKHDQDO\VHGSULQWLQJȴODPHQWVQDPHO\

Q\ORQDFU\ORQLWULOHEXWDGLHQHVW\UHQHKLJKLPSDFWSRO\VW\- rene, polyvinyl alcohol and polylactic acid, the latter re- leased the least TVOC ERs, and was the only one whose primary emitted monomer, lactide, was not considered a health risk²³7KXVDQHQFORVHGSULQWHUZLWKDQDLUȴOWUDWLRQ V\VWHPPD\EHUHFRPPHQGHG

In terms of environmental longevity, polylactide and OLJQLQEDVHGSRO\PHUDUHH[FHOOHQWSULQWLQJPDWHULDOVGXH WRWKHTXDQWLW\RIUHQHZDEOHUHVRXUFHVWKH\FRQWDLQ)XU- WKHUPRUHERWKȴODPHQWVDUHELRGHJUDGDEOHWRVRPHGH- JUHHLQGHHGWKHPDQXIDFWXUHUȇVVSHFLȴFDWLRQVVWDWHWKDW WKHOLJQLQEDVHGSRO\PHULVFRPSRVWDEOH^17,18, although no time span is indicated for this.

2YHUDOO)))SULQWLQJZLWKFRVWHɝFLHQWKLJKTXDOLW\DQG HQYLURQPHQWDOO\VXVWDLQDEOHSULQWLQJȴODPHQWVUHSUHVHQWV DQ LQJHQLRXV DGGLWLYH WHFKQRORJ\ WR EH XVHG LQ DOLJQHU orthodontics.

Conclusions

&RQVLGHULQJWKHOLPLWDWLRQVRIWKHV\VWHPVWXGLHGLWFDQEH FRQFOXGHGWKDWOD\HUKHLJKWD΍HFWVDFFXUDF\DQGSUHFLVLRQ EXWWKDWRWKHUSDUDPHWHUVVXFKDVSULQWLQJPDWHULDOVDQG VHWWLQJV LQȵXHQFH WKH UHVXOWV RI ))) SULQWLQJ $ KLJKHU Z-reso lution does not necessarily lead to higher accuracy DQGSUHFLVLRQUDWKHUWKHUHVHHPVWREHDQRSWLPXPUDQJH of layer heights depending on FFF print settings and ma- WHULDOΖQWKHSUHVHQWVWXG\WKHOLJQLQEDVHGSRO\PHUZDV VKRZQWREHDQH[FHOOHQW)))SULQWLQJPDWHULDOZLWKDQRS- timum layer height of 100.0 μm, even surpassing the precision requirements of the DLP printing control group.

)))SULQWLQJLVDKLJKTXDOLW\FRVWH΍HFWLYHDQGVXVWDLQ- DEOHWHFKQRORJ\IRUSURGXFLQJDOLJQHUPRGHOVZLWKUHVSHFW

to optimised layer height, print settings and material. In- deed, a higher layer height results in a higher printing vel ocity and thus exponentially shorter printing times )LJ 7DEOH 7KH RSWLPDO OD\HU KHLJKW ZLWK UHJDUG WR accuracy and precision in printing is approximately bwP)RU)))SULQWLQJDORZHUOD\HUKHLJKWR΍HUVQR DGYDQWDJHVLQWHUPVRIDFFXUDF\EXWUDWKHUOHDGVWRORQJ SULQWLQJ WLPHV DQG WKXV QRQHɝFLHQW SULQW ORDGV 8OWLP ately, a Z-resolution lower than 100.0 μm does not seem WR\LHOGDQ\HFRQRPLFRUFOLQLFDOEHQHȴW0RUHRYHUGHQWDO models printed using FFF with layer heights higher than wPVKRZEDUHO\DQ\ORVVRIDFFXUDF\ZLWKLQDFHUWDLQ UDQJHΖWZRXOGEHLQWHUHVWLQJWRLQYHVWLJDWHKRZWKHKLJK precision of FFF printed models correlates with the clinical HɝFDF\RIRUWKRGRQWLFDOLJQHUV7KXVIXWXUHVWXGLHVDUH UHTXLUHGWRGHWHUPLQHWKHPLQLPXPH΍HFWLYHOD\HUKHLJKW that transforms optimal forces onto the teeth using vacuum-formed aligners.

Declaration

7KHDXWKRUVGHFODUHWKHVHDUHQRFRQȵLFWVRILQWHUHVWUHODW- ing to this study.

References

1. Groth C, Kravitz ND, Jones PE, Graham JW, Redmond WR. Three- dimensional printing technology. J Clin Orthod 2014;48:475–485.

2. Pérez M, Medina-Sánchez G, García-Collado A, Gupta M, Carou D.

Surface quality enhancement of fused deposition modeling (FDM) SULQWHGVDPSOHVEDVHGRQWKHVHOHFWLRQRIFULWLFDOSULQWLQJSDUDPHWHUV 0DWHULDOV%DVHO

=KDQJ=&/L3/&KX)76KHQ*ΖQȵXHQFHRIWKHWKUHHGLPHQVLRQDO printing technique and printing layer thickness on model accuracy.

J Orofac Orthop 2019;80:194–204.

.DPLR 7 +D\DVKL . 2QGD 7 HW DO 8WLOL]LQJ D ORZFRVW GHVNWRS ' SULQWHUWRGHYHORSDȊRQHVWRS'SULQWLQJODEȋIRURUDODQGPD[LOOR IDFLDOVXUJHU\DQGGHQWLVWU\ȴHOGV'3ULQW0HG

+D]HYHOG$+XGGOHVWRQ6ODWHU--5HQ<$FFXUDF\DQGUHSURGXFLELOLW\

RIGHQWDOUHSOLFDPRGHOVUHFRQVWUXFWHGE\GL΍HUHQWUDSLGSURWRW\SLQJ techniques. Am J Orthod Dentofacial Orthop 2014;145:108–115.

%URZQ*%&XUULHU*).DGLRJOX2.LHUO-3$FFXUDF\RIGLPHQVLRQDO printed dental models reconstructed from digital intraoral impres- sions. Am J Orthod Dentofacial Orthop 2018;154:733–739.

'LHWULFK &$ (QGHU $ %DXPJDUWQHU 6 0HKO $ $ YDOLGDWLRQ VWXG\ RI UHFRQVWUXFWHG UDSLG SURWRW\SLQJ PRGHOV SURGXFHG E\ WZR WHFKQRO ogies. Angle Orthod 2017;87:782–787.

)DYHUR &6 (QJOLVK -' &R]DG %( :LUWKOLQ -2 6KRUW 00 .DVSHU ).

(΍HFW RI SULQW OD\HU KHLJKW DQG SULQWHU W\SH RQ WKH DFFXUDF\ RI 3-dimensional printed orthodontic models. Am J Orthod Dentofacial 2UWKRSȂ

(12)

/RȵLQ:$(QJOLVK-'%RUGHUV&HWDO(΍HFWRISULQWOD\HUKHLJKWRQ the assessment of 3D-printed models. Am J Orthod Dentofacial 2UWKRSȂ

10. Lee KY, Cho JW, Chang NY, et al. Accuracy of three-dimensional printing for manufacturing replica teeth. Korean J Orthod 2015;45:217–225.

&DPDUGHOOD/7%UHXQLQJ+9LOHOOD29$UHWKHUHGL΍HUHQFHVEHWZHHQ FRPSDULVRQPHWKRGVXVHGWRHYDOXDWHWKHDFFXUDF\DQGUHOLDELOLW\RI GLJLWDOPRGHOV"'HQWDO3UHVV-2UWKRGȂ

6KDKED]LDQ0-DFREV5:\DWW-HWDO$FFXUDF\DQGVXUJLFDOIHDVLELOLW\

RI D &%&7EDVHG VWHUHROLWKRJUDSKLF VXUJLFDO JXLGH DLGLQJ DXWRWUDQV- SODQWDWLRQRIWHHWKΖQYLWURYDOLGDWLRQ-2UDO5HKDELOȂ .UH\ .) 6FKLFNHU 3 6FKZDKQ & (LJHQZLOOLJ 3 %LRORJLVFKH 5HDOLW¦W FRPSXWHUJHSODQWHU =DKQEHZHJXQJHQ PLW $OLJQHUQ Ȃ HLQH PXOWL]HQ- trische, prospektive Untersuchung. Spitzenforschung in der Zahn- PHGL]LQȂ

%LEE5)UHHPDQ3%URZQ56XJDU$(YDQV3%RFFD$$QLQYHVWLJD- WLRQ RI WKUHHGLPHQVLRQDO VFDQQLQJ RI KXPDQ ERG\ VXUIDFHV DQG LWV use in the design and manufacture of prostheses. Proc Inst Mech Eng +Ȃ

15. Ryakhovskiy AN, Kostyukova VV. Comparative analysis of 3D data ac- FXUDF\ RI VLQJOH WRRWK DQG IXOO GHQWDO DUFK FDSWXUHG E\ GL΍HUHQW LQWUDRUDODQGODERUDWRU\GLJLWDOLPSUHVVLRQV\VWHPV>LQ5XVVLDQ@6WRP- DWRORJLLD0RVNȂ

&RUWHYLOOH ) .UH\ .) 5DW]PDQQ $ 2SWLPL]LQJ )))'SULQWLQJ IRU aligner orthodontics – Printing materials. IPJ 2020;4:2354.

([WUXGU *UHHQ7(&3UR 0DWHULDO 'DWD 6KHHW KWWSVZZZH[WUXGU FRPHQSURGXFWVFDWDORJXHJUHHQWHFSURQDWXUB $FFHVVHG

$SULO

([WUXGU 3/$ 1; 0DWHULDO 'DWD 6KHHW KWWSVZZZH[WUXGUFRPHQ SURGXFWVFDWDORJXHSODQ[RUDQJHB$FFHVVHG$SULO 19. Schirmer UR, Wiltshire WA. Manual and computer-aided space ana-

lysis: A comparative study. Am J Orthod Dentofac Orthop 1997;112:

Ȃ

+LURJDNL<6RKPXUD76DWRK+7DNDKDVKL-7DNDGD.&RPSOHWH' reconstruction of dental cast shape using perceptual grouping. IEEE Trans Med Imaging 2001;20:1093–1101.

+DOD]RQHWLV '- $FTXLVLWLRQ RI GLPHQVLRQDO VKDSHV IURP LPDJHV

$P-2UWKRG'HQWRIDFLDO2UWKRSȂ

&DVNR-69DGHQ-/.RNLFK9*HWDO2EMHFWLYHJUDGLQJV\VWHPIRUGHQ- WDOFDVWVDQGSDQRUDPLFUDGLRJUDSKV$PHULFDQ%RDUGRI2UWKRGRQ- tics. Am J Orthod Dentofacial Orthop 1998;114:589–599.

'DYLV$<=KDQJ4:RQJ-36:HEHU5-%ODFN06&KDUDFWHUL]DWLRQRI volatile organic compound emissions from consumer level material H[WUXVLRQ'SULQWHUV%XLOG(QYLURQȂ