The digital factory in both the modern dental lab and clinic · digital factory in both the modern dental lab and clinic David Leeson∗ Glidewell Dental, 4141 MacArthur Blvd, Newport

Page 1

Ta

DG

a

K

Z

C

I

M

D

1

Tinsti

h0

d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52

Available online at www.sciencedirect.com

ScienceDirect

jo ur nal home p ag e: www.int l .e lsev ierhea l th .com/ journa ls /dema

he digital factory in both the modern dental labnd clinic

avid Leeson ∗

lidewell Dental, 4141 MacArthur Blvd, Newport Beach, CA 92660, USA

r t i c l e i n f o

eywords:

irconia

AD/CAM

ntra oral scanning

icro CT scanning

ental devices

a b s t r a c t

Objective. Significant technological advances are occurring in the dental industry that are

complementary to the introduction of monolithic restorative materials. Glidewell Labora-

tories has been at the forefront of innovation of these developments and these internal

developments may be instructive of future trends in dentistry.

Methods. This paper examines internal Glidewell data from 2010 to 2019 about zirconia-

related technology and provides context for some technological advancements at Glidewell

Laboratories.

Results. The trend towards increased use of monolithic zirconia in both, posterior and ante-

rior regions continues to grow. With the advent of intraoral scanners and chairside CAD/CAM

technologies, more and more dentists are able to provide same day dentistry. Despite the

many clinicians that have made the leap to digital dentistry the low rate of growth in own-

ership and utilization means PVS impression will be common place long into the future.

Glidewell Laboratories has an advanced, automated impression scanning, AI crown design,

automated milling and glazing system extending the advantages of digital dentistry to all

dentists and pushing the bounds of the technology further.

Significance. This paper describes the first use of AI GAN to design dental crowns and also

shows the strong path monolithic zirconia restorations are paving into the future. Support-

ing technologies will continue to evolve around zirconia and this paper provides a snapshot

of those that exist and a few that are in development.

© 2019 The Academy of Dental Materials. Published by Elsevier Inc. All rights reserved.

support and complement the release of the material have beendeveloped over time. Glidewell Laboratories has developedand championed for several of these improvements and has

. Introduction

he dental industry has experienced unprecedented changen the last decade with rise of digital technologies and sig-ificant developments in restorative materials. The future iset to be even more transformative as the influence of Indus-

ry 4.0 [1] and Artificial Intelligence [2] influence all verticalsncluding dental.

∗ Correspondence to: 4141 MacArthur Blvd, Newport Beach, CA 92660, UE-mail address: [email protected]

ttps://doi.org/10.1016/j.dental.2019.10.010109-5641/© 2019 The Academy of Dental Materials. Published by Elsev

The introduction of monolithic zirconia (BruxZir® SolidZirconia, Glidewell Laboratories) in 2009 has been one suchmaterial that has influenced restorative dentistry with itsversatility and strength during use. Several technologies to

SA.

collected data over time of their use. Since Glidewell Laborato-

ier Inc. All rights reserved.

Page 2

44 d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52

Table 1 – TRAC research summarizes the key clinical points in current international standards for ceramics (ISO6872/ADA 69) [5].

Type of material Flexural strength (MPa) Example brand Indications/Fracture toughness (MPa/m0.5)

Class 5: Tetragonal Zirconia >800 MPa/KIc > 5 BruxZir Solid Zirconia Four or more units, anterior orposterior

Class 4: Cubic-ContainingZirconia

>500 MPa/KIc > 3.5 Lava Esthetic Three units, anterior orposterior

Class 3: LithiumDisilicate–High-StrengthGlass Ceramics

>300 MPa/KIc > 2.0 IPS e.max Single unit or three unitanterior

Class 2: LeuciteGlass-Ceramics

>100 MPa/KIc > 1.0

Class 1: Porcelains ≤100 MPa/KIc < 1.0

ries creates custom crowns, bridges and dentures for 70,000+of the 200,000 US Dentists [3], its internal data and new tech-nology development path can be instructive of wider trends inthe dental industry. This article specifically looks at this inter-nal data spanning the last decade or so to understand wherethe future of dentistry is headed.

2. Restorative materials

PFMs revolutionized dentistry with the ability to create strong,natural-looking restorations. However, with a flexural strengthof 80–100 MPa [4], the feldspathic porcelain often chips andcracks under use (Fig. 1).

Zirconia is chemically zirconium dioxide and is strongenough to create monolithic restorations that withstand sev-eral years of clinical service. Monolithic zirconia takes awaythe interface between metal and veneered ceramic that causesPFMs to fail. Table 1 is a summary of the strength of variousmaterials and indications for use.

With flexural strengths greater than 800 MPa [6], zirco-nia presented a strong and durable material for crowns andbridges, especially for use in patients with parafunctionalhabits. This change was reflected in the cases sent to GlidewellLaboratories. Fig. 2 compares the percentage breakdown of themajor restorative materials at Glidewell and charts the demiseof the PFM as the dominant material with the meteoric riseBruxZir as a monolithic Zirconia. In 2010, PFMs accounted for

50% of all cases sent to Glidewell compared to only 15% ofthe newly launched BruxZir. Veneered ceramics on zirconiaframeworks experienced a steady decline as did IPS e.max®

lithium disilicate after peaking in 2013. In 2019, PFM produc-

Fig. 1 – Chipped and cra

Ceramco Single unit, anterior orposterior, adhesively cementedVeneer ceramic, inlay ceramic

tion has seen a steady decline to 9% with BruxZir making up75% of all restorative cases sent to Glidewell Laboratories [7].

Fig. 3 shows the growth of BruxZir in more detail and a fore-cast using an exponential triple smoothing algorithm whichpredicts BruxZir will surpass 80% of restorative units orderedat Glidewell Laboratories by 2025. Clearly, monolithic zirconiawill continue to dominate the restorative market long into thefuture.

While zirconia can be used for posterior restorations, moretranslucency was needed for the anterior region. The translu-cency of zirconia can be modified by the addition of yttria tothe material. As the percentage of yttria in the formulationgoes up, the flexural strength of the resulting material goesdown. Higher translucency zirconia makes for a very life-likematerial that is well-suited for use in the anterior region ofthe mouth.

The evolution of the use of BruxZir by dentists can be betterunderstood through Fig. 4. The initial use of BruxZir is heavilybiased towards single units for the posterior region but as con-fidence in the use of the material grew and advancements inesthetics of the material through changes in composition andprocessing were made, the distribution of use is now closerto that of historical averages for PFMs noted in Fig. 5. Theintroduction of first generation higher yttria zirconia is clearin 2014 with a corresponding shift towards anterior units thefollowing year onwards [7].

Another factor of great significance is the evolving con-fidence in such products, which has been assisted byorganizations such as TRAC who have been performing clini-

cal testing of BruxZir since 2009. Technologies in RestorativesAnd Caries Research (TRAC Research) is a non-profit insti-tute dedicated to in-depth and long-term clinical studies of

cked PFM crowns.

Page 3

d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52 45

Fig. 2 – Restorative materials as a percentage of orders at Glidewell Laboratories [7].

Fig. 3 – BruxZir — growth to date and future growth prediction.

Fig. 4 – BruxZir usage.

Page 4

46 d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52

PFM usage.

Fig. 6 – Flexural strength (3 point bending per ISO

Fig. 5 –

restorative materials, preventive dentistry, and dental caries.This group studied a test group of 700+ BruxZir Solid zirconiarestorations that have a 100% survival at 7 years, making it themost durable restorative material tested in 43 years while sat-isfying many other needs including ease of use and minimalprep requirements. In fact, it was the only material that sat-isfied all seven requirements of an ideal restorative materialthat Christensen had outlined when they began this study [8].

The ease of use of BruxZir for the clinician is also reflectedin our own data having the lowest returns of any material inthe 49 year history of the company.

2.1. Improvements in zirconia technology

As previously stated, some of the growth of monolithic zirco-nia can be explained by improvements in the materials suchas improved translucency through the increase in yttria. How-ever, the increase in yttria results in a reduction in flexuralstrength and toughness [9]. Glidewell Laboratories developedand released Bruxzir Esthetic in 2018. The formulation andmanufacturing techniques for BruxZir EstheticTM are uniqueamongst higher yttria-containing zirconia in that they fur-ther improve translucency while maintaining strength andfracture toughness to allow BruxZir Esthetic to meet all indica-tions without compromise. BruxZir Esthetic defies the typicalstrength and translucency relationship with increasing yttriaas seen in Fig. 6. This is due to advances in the colloidal manu-facturing process that result in a more refined microstructurewhen compared to more common pressing techniques. Fig. 7has a representation of the two manufacturing processes col-loidal and pressing. The refined microstructure obtained usingthe colloidal manufacturing process results in better strengthand translucency.

3. Intraoral scanning

The previous decade has also seen significant growth in theuse of intraoral scanning in dental offices. When we lookedat the dentists sending their cases to Glidewell, we found that

6872:2015) vs translucency for different Zirconia materials.

26.44% of clinicians used an intraoral scanner in 2019 (Fig. 8)and if it follows the apparent linear growth of the prior decadethen by 2025 this will have increased to one third [7].

Looking deeper into scans received as a percentage of totalorders (Fig. 9), it is clear that utilization of the scanners hasimproved at much more rapid rate. This may reflect improvingtechnology that is faster and easier to integrate into practiceworkflows.

Isolating digital orders for BruxZir reveals a higher percent-age of orders from intraoral scans as seen in Fig. 6 shows thatBruxZir is leading the utilization trend. The rise of both mono-lithic Zirconia and the dramatic increase in scanner utilizationare certainly complementary. The ease of use of BruxZir andimproved accuracy enabled by a CAD/CAM manufacturing pro-cess, are likely additional drivers beyond the improvements inthe scanning technology itself.

Despite these improvements, the distribution of crownsand bridges for both digital orders and impressions, it is clearthat restorative digital orders are still skewed towards singleunits with a 33% bias vs PVS impression in 2019 at Glidewell.As demonstrated by Patzelt et al the additive effects of regis-

tration errors limit the cross arch accuracy accomplished withmany intra oral scanners, which may be one of the reasonsfor this bias [10]. There are also widely reported spatial limi-

Page 5

d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52 47

Fig. 7 – Uniaxial pressing (left) vs colloidal processing of ceramics right and associated micro structure.

Fig. 8 – Percentage of dentists that send their cases toGlidewell who own an intraoral scanner [7].

Fig. 9 – Intraoral scans received as a percentage of allorders at Glidewell compared to the percentage of intra orals

tmau

4

Iir

Fig. 10 – Automated Micro CT scanning for restorative cases

can BruxZir orders [7].

ations and other confounding factors for a given patient thatay drive the doctor to take a conventional PVS impression

s an alternative [11]. These factors are likely inhibiting fulltilization of the scanners.

. The digital factory at Glidewell

n order to meet the challenge of delivering consistent qual-ty at scale and satisfy the ever-growing demand for BruxZirestorations, Glidewell has made significant investment into

at Glidewell.

many patented and patent pending manufacturing technolo-gies. Beyond the speed and precision of these automatedmanufacturing systems, featured in Figs. 10 and 11, there arekey elements that can improve the dimensional accuracy andquality of the restorations produced.

4.1. Impression scanning

Given the slow rate of adoption of chairside intraoral scan-ning, physical impressions will remain the primary method ofacquiring the geometric information necessary for the fabri-cation of restorations for the foreseeable future.

In search of a method of offering every dentist a digitalworkflow, Glidewell has developed Industrial micro CT scan-ning technology. Micro CT scanning is well-proven in theorthodontic market and has been used for more than 17 years[12,13]. Glidewell is now pioneering its use for restorativeapplications and, as demonstrated by Kerr et al., can deliversuperior accuracy compared to conventional impressionspoured in stone over large spans [14]. It is therefore com-plementary to current intraoral scanning technology that haslimitations for full arch. The approach removes technician-dependent model pouring process variation and with an entire

case scanned in less than one minute, affords much faster pro-cessing times so restorations can be delivered to the dentistrapidly [14]. When combined with robotic automation, the CT

Page 6

48 d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52

Fig. 11 – Glidewell Digital Factory Major Process Steps (1) incoming PVS impressions are scanned using a micro-CT scanner(2) the digital micro-CT scan is used in the FastDesign (Glidewell Laboratories) software to design a restoration (3) therestoration is milled from a zirconia blank in the milling center (4) the restorations are manually pre-colored (5) and thensintered (6) all finished restorations undergo a scan that is then compared to the original design to ensure there are nodiscrepancies between planned and finished restoration (7) automated glazer then applies glaze and the restorations are

fired to bake the glaze.

scanner creates a pipeline that can be used to feed a digitalfactory alongside the scans received direct from practices.

4.2. AI in CAD design

Until recent advances at Glidewell, dental CAD software wasbased on predefined libraries that were premade for each

tooth. In the designing software, technicians chose a libraryto use for a certain case and then manually positioned andadjusted the predefined design to work with the specificpatient’s occlusion and adjacent teeth. This requires consid-

erable skill and the limitations of the predefined library setscan make it difficult to match the endless variation seen inhuman teeth.

Page 7

d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52 49

Fig. 12 – GAN generated crown design (right) vs library based (left) [7].

ge w

anLidstnetwGurulasga

4

Cai

Fig. 13 – Difference map for good vs defective brid

Glidewell Laboratories has the unique advantage of havingn archive of data from approximately 8 million cases withatural teeth and those designed by technicians. Glidewellaboratories used an artificial intelligence capable of learn-ng from these 8 million examples to be able to suggest betteresigns for crowns and bridges in the restoration designingoftware. Ian Goodfellow is widely considered to be the inven-or of Generative Adversarial Networks or GANs, where twoeural networks compete with each other in a game [15]. Forxample given a training data set of images one network syn-hesizes new images and the other network attempts to detecthich are fake and which are real. Glidewell’s application ofANs in collaboration with UC Berkeley is the first knownse for a manufactured product [16]. The technique createsestorations that are indistinguishable from the patient’s nat-ral anatomy as can be seen in Fig. 12. In older patients, the

ibrary-based method has a propensity to create more detailednatomy similar to a newly erupted tooth in contrast to theurrounding dentition which has wear from use. The GAN-enerated design replicates the natural wear patterns and theppropriate level of detail.

.3. Dimensional QC

onventionally, the only method of ensuring the dimensionalccuracy and therefore fit of a restoration, is by visually exam-ning it on a stone model before sending it to the clinician.

ith QC process flow chart (scale in micrometers).

This is subjective, based on the judgement of the evaluator,any wear that may happen on the stone model from pro-cessing, and cannot detect minute distortions. If a case ismanufactured from an intraoral scan, the only physical modelavailable would be generated by 3D printing, which itself issubject to manufacturing variability [17]. All of these issuesare solved with structured light optical scanning technologywhich captures a full 3D representation of the finished crownand compares it to the planned and designed model at themicron level as described in Fig. 13.

Due to the connected and automated nature of the manu-facturing system used to produce the crowns, feedback fromthe dimensional data is used to autonomously optimize thesystem, refining process parameters and improving the qual-ity of the resulting restorations. This is one of the key facetsof Industry 4.0, where aggregated data is used to create intel-ligent optimization of manufacturing processes [1].

4.4. Automated glazing

Glaze applied by conventional techniques such as brush orspray has inherent variability in thickness and texture, whichcompromises the accuracy of proximal contacts and cre-

ates delays because of the inevitable rework it necessitates.Repeated glazing also causes the shade of the crown to lightenand more seriously manual adjustment of the contacts risk theintroduction of sub surface damage that increases the proba-

Page 8

50 d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52

Fig. 14 – Automated Spray Glaze (left) v

Fig. 15 – Percentage of Glidewell Doctors that own a fullchairside CAD/CAM system [7].

BruxZir Solid Zirconia, maintaining the desirable mechan-

bility of fracture. To address this, a novel spray technology wascreated at Glidewell Laboratories, with a continuously con-trolled spray pattern that in combination with inputs fromCAD data, and by machine vision can apply a consistent glazeacross the entire surface of the restoration. With this tech-nology, an allowance for glaze thickness is designed into thecrown so the resulting proximal contacts are as designed oncethe glaze is applied. Fig. 14 shows an example of a conven-tionally spray glazed crown compared to the new technology.With the new technology the glaze can be controlled within a±5 �m range whereas conventional processes have variationof 50 �m or more.

4.5. Chairside CAD/CAM dentistry

Another trend worthy of examination is the expansion of fullchairside CAD/CAM systems across the last decade. There areseveral components in chairside CAD/CAM technology includ-ing an intra oral scanner, software to design the restorationand a digital means of fabrication, which today, is generally a

milling machine. The data in Fig. 15 is a subset of the scan-ner ownership percentages in Fig. 8. In 2019, approximately8.79% of Glidewell accounts owned a chairside CAD/CAM sys-

s Conventional Spray Glaze (right).

tem compared to 26.44% for overall scanner ownership andfollow broadly similar rates of growth (Fig. 16).

Considering, the relative proportions of types of restorativeproducts, doctors owning chairside systems send to Glidewell,is instructive of how the systems are employed in a typicalpractice. Predictably, the proportion of bridges relative to ante-rior and posterior single unit crowns is much higher than forthe average doctor and this balance remains relatively stablebetween 2010 and 2012. Less intuitive is the considerable shiftfrom 2013, with increasingly high proportions of single crownsbeing ordered throughout the decade. This corresponds to therise in BruxZir products and it is remarkable that the benefitsof monolithic zirconia were sufficient to influence the buyinghabits of doctors that have made substantial investments inchairside systems (Fig. 17).

4.6. glidewell.ioTM the digital factory in the clinic

Introduced in mid-2018, glidewell.ioTM In-Office Solution isa chairside CAD/CAM system including a scanner, millingmachine and design software. It makes use of the sameCAD/CAM technology and cloud infrastructure in a formappropriate for the clinic environment. The same AI technol-ogy that improves consistency in large scale manufacturingdelivers an automated less intensive design process toenhance the delivery of restorations in real time.

The system was envisioned to addresses the demand forZirconia observed from owners of existing chairside millingsystems in the form of BruxZir NowTM. BruxZir Now ismachined in the sintered state, which enables a crown to bedelivered in a single visit and negates the need for an oven.

BruxZir Now is the same formulation as the lab-made

ical properties that have made it so successful. The sameattributes that make it such a durable restorative material alsomake it one of the most challenging to produce. The system

Page 9

d e n t a l m a t e r i a l s 3 6 ( 2 0 2 0 ) 43–52 51

Fig. 16 – Break down of bridge, anterior and posterior crowns for Glidewell Doctors that own Chairside CAD/CAM Systemsexcluding glidewell.ioTM [7].

Fig. 17 – Break down of bridge, anterior and posterior crowns for Glidewell Doctors that own an intra-oral scanner [7].

Fig. 18 – SEM images of outer surface of lab produced pre sintered milled crown (left) and BruxZir Now sinter milled crown(right).

Page 10

l s 3

r

[17] Hazeveld A, Huddleston Slater JJR, Ren Y. Accuracy andreproducibility of dental replica models reconstructed bydifferent rapid prototyping techniques. Am J OrthodDentofacial Orthop 2014;145:108–15.

52 d e n t a l m a t e r i a

relies on numerous innovations in machine technology, CAMalgorithms and diamond abrasives to process this incrediblyhard and difficult to machine material with the bur remov-ing less than 10 microns of material per pass. This creates asurface texture even finer than a lab made crown that has apolished appearance and may be used directly without glaz-ing. SEM images comparing the surfaces of a lab manufacturedcrown with pre-sintered milling to a BruxZir Now machinedin sintered condition can be seen in Fig. 18.

5. Conclusion

Monolithic Zirconia is the single most important restora-tive material in the industry. The importance and suitabilityof monolithic zirconia to CAD/CAM processes lend itself tofurther technological advancements that result in improvedquality at scale and ever reducing cycle times when manufac-tured in the dental lab or the clinic:

• Despite the many clinicians that have made the leap todigital dentistry the low rate of growth in ownership andlagging utilization means PVS impression will be commonplace long into the future.

• Micro-CT scanning of a PVS impression can replace pouredstone models and is complimentary to intraoral scanninghaving improved accuracy over full arch scans.

• Machine learning techniques can be applied to dental CADand improve our ability to deliver more natural and con-sistent restorations. Machine learning is able to apply thelessons learned from millions of archived designs and realhuman dentition and incorporate them in the design ofevery crown produced.

• Digital technologies are also necessary to create a closedloop system of inspection that is independent of the subjec-tive judgment of a human. The resulting data sets are usedto autonomously optimize the system further improving thequality of restorations delivered.

• Monolithic zirconia crowns can now be milled in the clinicdelivering single visit crowns with a high polished appear-ance and excellent surface quality.

Acknowledgements

The author would like to thank Dr. Mayuri Kerr for exten-sive editing and writing assistance with this manuscript. Theauthor is also grateful to Ben Vu for providing access toGlidewell Laboratories internal data.

e f e r e n c e s

[1] Cotteleer M. How industrial IoT will reshape manufacturing

— and the bottom line. Harv Bus Rev 2017, . [Accessed 3September 2019]https://hbr.org/webinar/2017/04/how-industrial-iot-will-reshape-manufacturing-and-the-bottom-line.

6 ( 2 0 2 0 ) 43–52

[2] Davenport TH, Ronanki R. 3 things AI can already do foryour company. Harv Bus Rev 2018:108–16.

[3] ADA Health Policy Institute FAQ — Dental Workforce in theU.S. n.d. https://www.ada.org/en/science-research/health-policy-institute/dental-statistics/workforce.[Accessed 3 July 2019].

[4] Guazzato M, Proos K, Quach L, Swain MV. Strength, reliabilityand mode of fracture of bilayered porcelain/zirconia (Y-TZP)dental ceramics. Biomaterials 2004;25:5045–52.

[5] Christensen R. Ceramic crowns: where are we in 2019?; 2019.p. 90–7 Https://WwwDentaltownCom/Magazine/Article/7576/Aesthetic-Ceramic-Crowns-Where-Are-We-in-2019.

[6] Filser F, Kocher P, Weibel F, Lüthy H, Schärer P, Gauckler LJ.Reliability and strength of all-ceramic dental restorationsfabricated by direct ceramic machining (DCM). Int J ComputDent 2001;4:89–106.

[7] Glidewell Laboratories. Glidewell dental internal data; 2019.[8] Christensen G. Zirconia: most durable tooth-colored crown

material in practice-based clinical study. Clin Rep2018;11:3–6.

[9] Kwon SJ, Lawson NC, McLaren EE, Nejat AH, Burgess JO.Comparison of the mechanical properties of translucentzirconia and lithium disilicate. J Prosthet Dent2018;120:132–7.

[10] Patzelt SBM, Emmanouilidi A, Stampf S, Strub JR, Att W.Accuracy of full-arch scans using intraoral scanners. ClinOral Investig 2014;18:1687–94.

[11] Keeling A, Wu J, Ferrari M. Confounding factors affecting themarginal quality of an intra-oral scan. J Dent 2017;59:33–40.

[12] Steinhäuser-Andresen S, Detterbeck A, Funk C, Krumm M,Kasperl S, Holst A, et al. Pilotstudie zur Genauigkeit undDimensionsstabilität von Abformmaterialien mittelsindustrieller CT-Messungen. J Orofac Orthop 2011;72:111–24.

[13] Align Enhances Manufacturing Process with HYTEC’sFlashCT Computed Tomography Scanners|Align Technologyn.d. http://investor.aligntech.com/news-releases/news-release-details/align-enhances-manufacturing-process-hytecs-flashct-computed. [Accessed 4 July 2019].

[14] Kerr M, Park N, Leeson D, Nikolskiy S. Dimensional accuracyof microcomputed tomography-scanned half-archimpressions. J Prosthet Dent 2019;121:797–802.

[15] Goodfellow I, Pouget-Abadie J, Mirza M, Xu B, Warde-FarleyD, Ozair S, et al. Generative adversarial nets. Adv Neural InfProcess Syst 2014:2672–80.

[16] Hwang JJ, Azernikov S, Efros AA, Yu SX. Learning beyondhuman expertise with generative models for dentalrestorations; 2018, arXiv preprint arXiv:1804.00064.