Abstract
PURPOSE
This study examines the potential of computer-aided design (CAD) systems equipped with artificial intelligence (AI) in reducing the workload of dental technicians. We aimed to compare the accuracy and design time of crowns designed using conventional CAD with those designed using AI-equipped CAD.
MATERIALS AND METHODS
Abutment tooth models of a maxillary right second premolar (FDI classification #15) and a maxillary left first molar (FDI classification #26) were mounted on a dental model to form the master model. Stereolithography data were acquired using an intraoral scanner, and five dental technicians designed one crown each for #15 and #26 using both conventional and AI-equipped CAD systems. With the #15 and #26 crowns, six measuring points were established for comparing the accuracy of the occlusal surfaces and design time of the crowns designed by the two CAD systems. The occlusal surfaces were also compared for the buccal and palatal sides.
RESULTS
The accuracy of the occlusal surface was 275.5 ± 116.8 µm and 25.7 ± 13 µm for the conventional CAD and AI-equipped CAD systems, respectively. For the buccal and palatal surface comparisons, the conventional CAD system revealed larger misfits on the palatal side for both #15 and #26, with significant differences observed. No significant differences were noted with the AI-equipped CAD system. The AI-equipped CAD resulted in significantly faster design time for both #15 and #26.
CONCLUSION
The AI-based CAD system significantly reduced design time and enabled the fabrication of uniform crowns regardless of the dental technician's experience and skill.
Keywords: Artificial intelligence, Computer-aided design, Crowns, Dental models
INTRODUCTION
Recent advancements in digital technology have significantly benefited dental treatment, aiding dentists, dental technicians, and patients alike.1,2,3 Traditional crown fabrication involves taking impressions using alginate, agar, or silicone, injecting plaster for creating models, and casting waxed-up designs.4 Modern prosthodontics leverages intraoral scanners (IOS) and computer-aided design-computer-aided manufacturing (CAD-CAM) systems for fabricating prosthetics digitally.5,6 Additionally, three-dimensional (3D) printers have become common substitutes for plaster casts.7 Digitization in dental treatment has contributed to a shorter prosthetic fabrication time, improved accuracy, and reduced workload for the dental technicians.8,9 However, the Japanese Ministry of Health, Labor, and Welfare reported that the number of employed dental technicians in FY2020 was approximately 35,000 with 52 % being > 50 years old. With an overall rate of change of only 1 %, the number of dental technicians is expected to decline in the future.10 This decline in the number of dental technicians may affect dental care. Artificial Intelligence (AI), a technology currently attracting attention, can improve the quality and accuracy of medical care at low cost.11 However, few reports have explored the use of AI in dental treatment, with existing research focusing on implant recognition and tooth shade accuracy.12,13 In prosthetic treatment, Bernauer's systematic review reported that of the seven articles screened, six reported training and application of AI systems, with no published clinical reports.14 Similarly, the systematic review of Ali et al. indicated that the research focus was primarily on the evaluation at the diagnostic stage, with few studies on treatment planning and implementation.15 Recently, AI-equipped CAD systems have been developed16; however, reports on their accuracy are limited. We believe that highly fitting CAD-CAM crowns can be produced more quickly by using CAD equipped with AI, regardless of the dental technician's experience or skill, thereby reducing the burden on dental technicians. Therefore, this study aimed to compare the accuracy and design time of crowns designed using conventional CAD systems and those designed using AI-equipped CAD systems.
MATERIALS AND METHODS
Ethical approval was not required by the Kanagawa Dental College as this study used models. Separate abutment tooth models (A55A-152, A55A-262; NISSIN, Tokyo, Japan) of the maxillary right second premolar (FDI classification #15) and maxillary left first molar (FDI classification #26) were mounted on the dental model ([D16FE-500A(GSE)-QF], NISSIN, Tokyo, Japan) to serve as the master model (Fig. 1). Digital impressions of the master model were made using an IOS (Trios 3®; 3Shape, Copenhagen, Denmark), and stereolithography (STL) data were obtained. Based on the STL data, crowns for #15 and #26 were designed using both conventional CAD (3Shape Dental System®; 3Shape, Copenhagen, Denmark) and AI-equipped CAD (Dentbird; Imagoworks Inc., Seoul, Korea). The design parameters were set as follows: margin width, 80 µm; margin angle, 50°; occlusal distance, 0 µm; approximal distance, −20 µm; cement gap, 40 µm; and adaptive extra gap, 50 µm based on the design of Cho et al.17 (Fig. 2). The same parameters were applied to both the 3Shape Dental System® and Dentbird system. For the 3Shape Dental System®, the 3Shape Smile Library was selected with “Supranatural by Lorant Stumpf” for #15 and “Allure DS2 RealView” for #26. Crowns were milled from a CAD-CAM block (Cerasmart-prime; GC Japan, Tokyo, Japan) using a DWX-53DC (DENKEN-HIGHDENTAL Co., Ltd., Kyoto, Japan). Crowns designed with the 3Shape Dental System® are referred to as Cr1, and crowns designed by Dentbird are referred to as Cr2. As milling conditions, bar sizes ZPB-100D R1.0 and ZPB-50D R0.5 (DGSHAPE Corporation, Shizuoka, Japan) were used; stepover was set at 0.2 – 0.5 mm and 0.05 – 0.1 mm in XY, respectively. The stage feed rate was set at 2,000 mm/min and 800 mm/min, respectively, and the spindle speed was 150,000 rpm and 250,000 rpm, respectively. Five dental technicians, with 32, 23, 19, 16, and 3 years of experience, respectively, each designed one crown for #15 and #26 using both CAD systems and fabricated a total of 20 crowns. The Dentbird system's process involved the following steps: selecting “start a new case,” dragging and dropping the STL file, selecting “Crown,” entering the design parameters, and selecting “Generate Crown,” thereby allowing AI to automatically design the case (Fig. 3). In this study, the accuracy of the two CAD systems, the intensity of proximal contact after modeling, design time, and marginal fit were measured and compared. All the study items were measured by one dentist.
Fig. 1. (A) #15 abutment tooth models were mounted on a dental model and used as master models, (B) #26 abutment tooth models were mounted on a dental model and used as master models.
Fig. 2. Description of the parameters during design. (A) Details of margin width, margin angle, cement gap, and adaptive extra gap, (B) Detail of occlusal distance, (C) Detail of approximal distance. Figure adapted with permission from Imagoworks Inc. (https://cloud.dentbird.com/design).
Fig. 3. Dentbird Design Procedure. (A) Selecting a design file, (B) Upper and lower jaw selection, (C) Auto-detection of margin lines, (D) Automatic crown design.
Regarding accuracy, after milling, the crowns were placed on the master model without cementation, and STL data were obtained using a 3Shape E3 3D scanner (3Shape, Copenhagen, Denmark). The data were then superimposed using Geomagic® Control (3D Systems, Washington, DC, USA). Measurement points a-f were distinguished between Cr1 and Cr2, and the errors were measured among the dental technicians. For instance, for both Cr1 and Cr2, the comparisons were conducted between dental technician 1-2, 1-3, and 2-3. For superimposition, the bestfit alignment tool was selected and the dentition was superimposed. Using Geomagic Control X, the root mean square (RMS) estimates of the two data sets were measured, with the misfit of dental technician 1-2 as the reference, the difference in misfit per dental technician as the trueness, and the means of all as the accuracy. In this case, trueness was set as accuracy.18 The accuracy of the occlusal surface was measured by setting the buccal cusp of # 15 as point a, palatal cusp as point b, mesial buccal cusp of # 26 as point c, distal buccal cusp as point d, mesial palatal cusp as point e, and distal palatal cusp as point f (Fig. 4). Accuracy was measured automatically and only one measurement was taken. Cr1 and Cr2 misfits at #15 and #26 were compared for each measurement site. A t-test was used to compare the accuracy of each measurement site. Statistical analyses were done using Bell-Curve for Excel (Social Survey Research Information Co., Ltd., Tokyo, Japan). The differences with P value < .05 were considered statistically significant. Additionally, we examined whether there was a difference between the buccal and palatal sides of #15 and #26. In other words, we distinguished the measurement points a, c and d, and b, e and f, and compared them using a t-test. The differences with P values < .05 were considered statistically significant.
Fig. 4. (A) The buccal cusp of #15 was set as a and the palatal cusp as b, (B) The mesial-buccal cusp of #26 was set as c, the distal-buccal cusp as d, the mesial-palatal cusp as e, and the distal-palatal cusp as f.
Regarding the strength of the proximal contact between the milled crown and adjacent teeth, the crown was placed on a model and the proximal contact of the adjacent teeth was measured at 40 locations. The crowns were fitted to a model, and the proximal contacts of the adjacent teeth were measured at 40 locations. The crowns were divided into three groups: one group with dental floss (REACH® Classic Clean Floss; LG H&H Co., Ltd., Seoul, Korea) (Floss group), one group with a contact gauge (GC, Tokyo, Japan) of 50 µm (50 µm group), and one group with a contact gauge of 110 µm (110 µm group). Cr1 fabricated by CAD equipped with AI showed better contact with the adjacent teeth.
Regarding design time, each dental technician compared crown design time using the 3Shape Dental System® and Dentbird. The design times from “start a new case” to the end of crown design for Dentbird and from “new order” to the end of crown design for 3Shape Dental System® were measured. The design times of the two CAD systems for each tooth type were compared using a t-test. The differences with P values < .05 were considered statistically significant.
Regarding the Cr1 and Cr2 marginal fit, the crowns were placed on the abutment tooth without intervention and aligned perpendicular to the x-ray beam in the micro computed tomography (CT) tube. Imaging was performed using micro-CT (ScanXmate-L080T; Comscantecno Co., Ltd., Kanagawa, Japan) under the following imaging conditions: 60 kV, 80 µA, and 2.891 × magnification. After imaging, the marginal fit of the abutment and crown was measured using the 3D viewer function of the 3D image analysis system volume analyzer (SYNAPSE VINCENT®; Fujifilm, Tokyo, Japan). Measurements were performed at four points around the tooth: the mesial center (M), distal center (D), buccal center (B), and palatal center (P) (Fig. 5).19
Fig. 5. (A) Measurement site for #15 marginal fit, (B) Measurement site for #26 marginal fit.
RESULTS
The accuracy of Cr1 was 198.1 ± 76.7 µm for a, 362.7 ± 177.3 µm for b, 153.6 ± 96.3 µm for c, 172.4 ± 74.2 µm for d, 435.2 ± 107.4 µm for e, and 331.1 ± 169.1 µm for f. The accuracy of Cr2 was 24.0 ± 8.4 µm for a, 22.2 ± 9.3 µm for b, 29.3 ± 13.6 µm for c, 27.8 ± 17 µm for d, 23.0 ± 16.3 µm for e, and 28.2 ± 13.4 µm for f (Table 1). Significant differences were found at all measurement points (Table 2). In the buccopalatal comparison, significant differences were observed in Cr1 for both #15 and #26, with larger errors on the palatal side (b, e, and f). No significant differences were observed for Cr2. (P < .01) (Table 3).
Table 1. Misfit among dental technicians at each measurement point.
| a | b | c | d | e | f | ||
|---|---|---|---|---|---|---|---|
| Cr1 | DT1-DT2 | 157.2 | 531.3 | 163.3 | 233.5 | 425.1 | 581.8 |
| DT1-DT3 | 137.6 | 386.3 | 65.6 | 117.4 | 523.3 | 283.7 | |
| DT1-DT4 | 223.4 | 306.8 | 80.9 | 122.3 | 368.6 | 289.5 | |
| DT1-DT5 | 220.6 | 259.3 | 142.8 | 189.2 | 574.9 | 347.6 | |
| DT2-DT3 | 134.3 | 531.2 | 116.8 | 224.3 | 364.3 | 238.5 | |
| DT2-DT4 | 216.7 | 632 | 158 | 329 | 273 | 116 | |
| DT2-DT5 | 250.8 | 469.3 | 161.7 | 155.9 | 476.8 | 554.9 | |
| DT3-DT4 | 373.3 | 168.5 | 161.3 | 106.4 | 289.6 | 190.6 | |
| DT3-DT5 | 142.1 | 79.4 | 405.3 | 82.4 | 514.4 | 533.8 | |
| DT4-DT5 | 124.8 | 262.6 | 80.3 | 163.4 | 542.3 | 174.5 | |
| means | 198.1 | 362.7 | 153.6 | 172.4 | 435.2 | 331.1 | |
| standard deviations | 76.7 | 177.3 | 96.3 | 74.2 | 107.4 | 169.1 | |
| Cr2 | DT1-DT2 | 24.6 | 16.9 | 31 | 16 | 19.1 | 40.1 |
| DT1-DT3 | 22.4 | 45.7 | 27.7 | 56.2 | 41.1 | 28.4 | |
| DT1-DT4 | 21.6 | 18 | 23.2 | 30.7 | 10.8 | 38.4 | |
| DT1-DT5 | 15.1 | 20 | 31.1 | 56 | 22.9 | 13 | |
| DT2-DT3 | 18.5 | 20.3 | 39.9 | 22.4 | 23.1 | 14.9 | |
| DT2-DT4 | 25.3 | 19.7 | 33 | 38 | 10 | 12.6 | |
| DT2-DT5 | 21.2 | 30.4 | 16.9 | 14.6 | 9.2 | 14.2 | |
| DT3-DT4 | 30.5 | 14.4 | 59.4 | 16.6 | 60.7 | 47.1 | |
| DT3-DT5 | 16 | 17.7 | 20.2 | 17.2 | 12 | 39.4 | |
| DT4-DT5 | 44.3 | 19 | 10.3 | 10.3 | 21.1 | 33.6 | |
| means | 24 | 22.2 | 29.3 | 27.8 | 23 | 28.2 | |
| standard deviations | 8.4 | 9.3 | 13.6 | 17 | 16.3 | 13.4 |
The misfit of dental technician 1 - 2 was used as the standard, and the difference in misfit between dental technicians was defined as trueness.
DT: dental technicians.
Table 2. Comparison of the accuracy of each measurement point (P < .05).
| a (µm) | b (µm) | c (µm) | d (µm) | e (µm) | f (µm) | means (µm) | |
|---|---|---|---|---|---|---|---|
| Cr1 | 198.1 ± 76.7 | 362.7 ± 177.3 | 153.6 ± 96.3 | 172.4 ± 74.2 | 435.2 ± 107.4 | 331.1 ± 169.1 | 275.5 ± 116.8 |
| Cr2 | 24.0 ± 8.4 | 22.2 ± 9.3 | 29.3 ± 13.6 | 27.8 ± 17 | 23.0 ± 16.3 | 28.2 ± 13.4 | 25.7 ± 13 |
| Significant difference | P < .001 | P < .001 | P < .001 | P < .001 | P < .001 | P < .001 |
The mean value of trueness was used to represent accuracy.
Table 3. Comparison of the buccal and palatal accuracy (P < .05).
| Cr1 | Cr2 | ||
|---|---|---|---|
| Comparison of a and b. | a (µm) | 198.1 ± 76.7 | 24.0 ± 8.4 |
| b (µm) | 362.7 ± 177.3 | 22.2 ± 9.3 | |
| significant difference | P = .01 | P = .67 | |
| Comparison of c, d and e, f. | c, d (µm) | 163.0 ± 84.2 | 28.5 ± 15 |
| e, f (µm) | 383.2 ± 147.9 | 25.6 ± 14.7 | |
| Significant difference | P < .01 | P = .54 |
The intensity of proximal contacts between adjacent teeth was 40%, 50%, and 10% for the floss, 50-µm, and 110-µm groups, respectively, in Cr1, while 70% and 30% for the floss and 50-µm groups, respectively, in Cr2. The 110-µm group was not observed in Cr2 (Table 4).
Table 4. Results of the intensity of proximal contact with Cr2 showing better contact.
| Floss group | 50-µm group | 110-µm group | |
|---|---|---|---|
| Cr1 | 8 (40%) | 10 (50%) | 2 (10%) |
| Cr2 | 14 (70%) | 6 (30%) | 0 |
The design time for #15 was 397.2 ± 80.4 min for the 3Shape Dental System® and 99.4 ± 17.1 min for Dentbird. The design time for #26 was 516.4 ± 61.3 min for the 3Shape Dental System® and 97.6 ± 11.1 min for Dentbird. Both #15 and #26 showed significant differences (P < .001) (Table 5).
Table 5. Comparison of the design time between the 3Shape Dental System® and Dentbird.
| #15 (min) | #26 (min) | |
|---|---|---|
| 3Shape Dental System® | 397.2 ± 80.4 | 516.4 ± 61.3 |
| Dentbird | 99.4 ± 17.1 | 97.6 ± 11.1 |
| Significant difference | P < .001 | P < .001 |
Marginal fits of Cr1 in #15 was 52 ± 13 µm for M, 56 ± 15 µm for D, 66 ± 11 µm for B, and 54 ± 17 µm for P. Cr2 was 56 ± 11 µm for M, 62 ± 8 µm for D, 72 ± 8 µm for B, and 62 ± 19 µm for P. The marginal fits of Cr1 in #26 were 68 ± 13 µm for M, 64 ± 9 µm for D, 72 ± 11 µm for B, and 68 ± 16 µm for P. For Cr2, M was 60 ± 12 µm, D was 68 ± 17 µm, B was 76 ± 11 µm, and P was 74 ± 11 µm. No significant difference was observed for #15 and #26 (Table 6).
Table 6. Comparison of Cr1 and Cr2 marginal fits.
| M (µm) | D (µm) | B (µm) | P (µm) | ||
|---|---|---|---|---|---|
| #15 | Cr1 | 52 ± 13 | 56 ± 15 | 66 ± 11 | 54 ± 17 |
| Cr2 | 56 ± 11 | 62 ± 8 | 72 ± 8 | 62 ± 19 | |
| Significant difference | P = .62 | P = .46 | P = .37 | P = .50 | |
| #26 | Cr1 | 68 ± 13 | 64 ± 9 | 72 ± 11 | 68 ± 16 |
| Cr2 | 60 ± 12 | 68 ± 17 | 76 ± 11 | 74 ± 11 | |
| Significant difference | P = .35 | P = .62 | P = .59 | P = .52 |
DISCUSSION
This study compared the accuracy and design time of crowns designed using conventional CAD and those designed using AI-equipped CAD. The AI-equipped CAD made it possible to design and fabricate crowns with good accuracy in a short time. Dentbird uses deep learning technology to automatically recognize abutment teeth and margin lines using AI's positioning function and can complete the data processing in a few clicks. Additionally, because the design can be performed online, no CAD software is required, making the system easily accessible and cost effective. Several reports have compared Dentbird, an AI-equipped CAD system, with conventional CAD systems.
Regarding occlusal accuracy, Cho et al.20 measured the difference in occlusal morphology in crown design as we did; they reported that the 3Shape Dental System had an error of 270.4 ± 72.4 µm and Dentbird had an error of 49.6 ± 19.6 µm. Cho et al.17 also compared the occlusal contacts between Dentbird and Dental System. In the analysis of the number of occlusal contact points, they reported that the Dental System had the highest mean number of occlusal contact points per designed crown at 5.5 ± 1.5, while the Dentbird had 4.3 ± 1.7, showing a significant difference. In this study, the accuracy of the occlusal surfaces was measured at positions a to f. The mean values of Cr1 and Cr2 were 275.5 ± 116.8 µm and 25.7 ± 13 µm, respectively, similar to those reported by Cho et al. We compared the results by distinguishing the measurement points between the buccal and palatal sides, and found that the misfit of b, e, and f, which are on the palatal side in Cr1, are significantly larger than those of b, e, and f. The palatal side of the maxilla is a functional occlusion head, which indicates that dental technicians have different experiences, skills, and ideas about occlusion assignment. However, no significant difference was found in Cr2. Considering the reports on occlusal contact by Cho and others, CAD equipped AI is capable of providing uniform occlusal contact.
Regarding design time, Cho et al.20 measured the time efficiency of crown design and the difference in occlusal morphology of virtual crowns. The total elapsed time for the entire crown design and optimization process was 371.7 ± 89.0 min for the 3Shape Dental System and 267.1 ± 91.5 min for Dentbird, with Dentbird taking less time and showing a significant difference. Regarding the design time, Dentbird took less time than 3Shape Dental System® for both #15 and #26, showing a significant difference, which is consistent with their report.
Regarding fit accuracy, Choi et al.21 designed crowns to compare the accuracy of Dentbird, 3Shape, Exocad, and MEDIT. They measured the Hausdorff distance (HD) and chamfer distance (CD) from the position on the abutment to the finish line to verify the accuracy of the finish line of the crown. The threshold of HD for the 51 abutment teeth in the IOS group was 0.566 mm and the threshold of CD was 0.100 a.u. The HD of Dentbird was 0.543 ± 0.274 mm, and the CD was 0.089 ± 0.076 a.u.; HD of 3Shape was 0.712 ± 0.325 mm, and CD was 0.17 ± 0.157 a.u.; HD of Exocad was 0.635 ± 0.332 mm, and CD was 0.124 ± 0.131 a.u.; and HD of MEDIT was 0.694 ± 0.372 mm, and CD was 0.16 ± 0.209 a.u. Only in Dentbird, the mean values of CD and HD were below the threshold, and the finish line was clinically acceptable. Dentbird also used a convolutional neural network (CNN), and reported that 12 of the 168 abutment teeth in the clinical data could not be detected; however, 93% of the total teeth could be detected. Rençber Kızılkaya et al.22 designed Dentbird and Exocad with Inlab 20, fabricated temporary crowns from a 3D printer, and measured the marginal fit. The results of superimposing the master model with Geomagic Control X showed that the B of Dentbird was -0.04 ± 0.08 µm, P was 0.08 ± 0.04 µm, M was 0.16 ± 0.22 µm, D was 0.38 ± 0.14 µm, B of Exocad was 0.12 ± 0.11 µm, P was 0.12 ± 0.11 µm, and D was 0.38 ± 0.14 µm; B of Inlab 20 was 0.03 ± 0.07 µm, P was 0.01 ± 0.09 µm, M was 0.31 ± 0.1 µm, and D was 0.55 ± 0.12 µm. Dentbird reported the best fit. Thus, the detection rate of Dentbird for abutment teeth is high, with good reported accuracy. Aktaş et al.23 verified the accuracy of marginal fit of crowns designed using Dentbird and Dentalcad. The abutment tooth model #46 was mounted on the master model, crowns were designed using Dentbird and Dentalcad, and 10 crowns were fabricated for each design using a 3D printer. After the crowns were fabricated, they were fit to the models and measured by micro-CT. The results were 60 ± 32 µm in the Dentbird 3D printer group, 75 ± 28 µm in the Milled group, 103 ± 30 µm in the Dentalcad 3D printer group, and 54 ± 43 µm in the Milled group. No significant differences were reported in the Milled group. There are many reports on the use of micro-CT to measure the marginal fit of prosthetic devices. Daou et al.24 reported that the marginal gap between the abutment teeth of zirconia 3-unit fixed dental prostheses and typodont mode was 30.95 ± 6.17 µm and 40.42 ± 9.58 µm, respectively. For the marginal fit between crown and abutment teeth, ≤ 120 µm is desirable.25,26,27 The results of our study item 4 showed that both Cr1 and Cr2 of #15 and 26 are less than 120 µm, which is similar to the results of the Milled group of the study by Aktaş et al.
No study has verified the accuracy of the design of the same abutment tooth by multiple dental technicians using AI-equipped CAD and conventional CAD. Dentbird is not dependent on the experience and skill of the dental technician, and it is suggested that Dentbird can shorten the design time and create crowns with a good fit. This will enable the uniform creation of prosthetics, improve work efficiency, and reduce the workload of dental technicians as the number of dental technicians is decreasing. Lin et al.28 examined dental technicians' perceptions of AI in the context of their use of AI. The results of interviews with 12 subjects affirmed the potential of AI to improve efficiency, accuracy, and communication in dental laboratory work. However, as a disadvantage, they also reported that concerns were raised regarding employment stability, ethical considerations, and the need for training and support. Conventional dental treatment consists of dentists, dental technicians, dental hygienists, and patients; but from now on, it is necessary to establish a new environment with the addition of AI.
As this is a model study, the margin lines are considered clear. In clinical practice, the margin line detection time of AI may differ depending on the gingival condition and whether the margin line is above or below the gingival margin. Moreover, as mentioned earlier, as cement was not used in this study, no lifting was caused by cement, which may have contributed to the good fit. Therefore, it may be necessary to measure the accuracy and adjust the parameters of the design in clinical practice in the future. Previous studies reported that each patient's dentition and oral prosthetic material is different.29 Additionally, no sample size measurements were considered in this study and the number of crowns was small; therefore, it is necessary to distinguish between the oral conditions and measure sample sizes in a clinical setting to verify accuracy in the future.
CONCLUSION
In this study, crowns were designed using both the 3Shape Dental System®, a conventional CAD system, and Dentbird, an AI-equipped CAD system. The crowns designed by Dentbird minimized the errors caused by the craftsman's technique, and a good fit was achieved in terms of occlusal surface, contact strength of adjacent teeth, and marginal fit. In addition, the time required for crown design was reduced. Therefore, we believe that Dentbird can improve crown accuracy and, thereby, reduce the workload of dental technicians.
References
- 1.Lo Russo L, Caradonna G, Biancardino M, De Lillo A, Troiano G, Guida L. Digital versus conventional workflow for the fabrication of multiunit fixed prostheses: a systematic review and meta-analysis of vertical marginal fit in controlled in vitro studies. J Prosthet Dent. 2019;122:435–440. doi: 10.1016/j.prosdent.2018.12.001. [DOI] [PubMed] [Google Scholar]
- 2.Aktas G, Özcan N, Aydin DH, Şahin E, Akça K. Effect of digitizing techniques on the fit of implant-retained crowns with different antirotational abutment features. J Prosthet Dent. 2014;111:367–372. doi: 10.1016/j.prosdent.2013.11.001. [DOI] [PubMed] [Google Scholar]
- 3.Nagata K, Fuchigami K, Okuhama Y, Wakamori K, Tsuruoka H, Nakashizu T, Hoshi N, Atsumi M, Kimoto K, Kawana H. Comparison of digital and silicone impressions for single-tooth implants and two- and three-unit implants for a free-end edentulous saddle. BMC Oral Health. 2021;21:464. doi: 10.1186/s12903-021-01836-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hansson O, Eklund J. A historical review of hydrocolloids and an investigation of the dimensional accuracy of the new alginates for crown and bridge impressions when using stock trays. Swed Dent J. 1984;8:81–95. [PubMed] [Google Scholar]
- 5.Kernen F, Schlager S, Seidel Alvarez V, Mehrhof J, Vach K, Kohal R, Nelson K, Flügge T. Accuracy of intraoral scans: an in vivo study of different scanning devices. J Prosthet Dent. 2022;128:1303–1309. doi: 10.1016/j.prosdent.2021.03.007. [DOI] [PubMed] [Google Scholar]
- 6.Zhou Y, Fu L, Zhang Z, Tang X. Effect of tooth color on the accuracy of intraoral complete arch scanning under different light conditions using a zirconia restoration model. J Prosthet Dent. 2024;131:145.e1–145.e8. doi: 10.1016/j.prosdent.2023.10.001. [DOI] [PubMed] [Google Scholar]
- 7.Wakamori K, Nagata K, Nakashizu T, Tsuruoka H, Atsumi M, Kawana H. Comparative verification of the accuracy of implant models made of PLA, resin, and silicone. Materials (Basel) 2023;16:3307. doi: 10.3390/ma16093307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Joda T, Lenherr P, Dedem P, Kovaltschuk I, Bragger U, Zitzmann NU. Time efficiency, difficulty, and operator’s preference comparing digital and conventional implant impressions: a randomized controlled trial. Clin Oral Implants Res. 2017;28:1318–1323. doi: 10.1111/clr.12982. [DOI] [PubMed] [Google Scholar]
- 9.Peroz S, Spies BC, Adali U, Beuer F, Wesemann C. Measured accuracy of intraoral scanners is highly dependent on methodical factors. J Prosthodont Res. 2022;66:318–325. doi: 10.2186/jpr.JPR_D_21_00023. [DOI] [PubMed] [Google Scholar]
- 10.Oshima K. Current status of supply of and demand for dental technicians in Japan: evaluation and countermeasures against the decrease in the number of dental technicians. Jpn Dent Sci Rev. 2021;57:123–127. doi: 10.1016/j.jdsr.2021.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Agrawal P, Nikhade P. Artificial intelligence in dentistry: past, present, and future. Cureus. 2022;14:e27405. doi: 10.7759/cureus.27405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Revilla-León M, Gómez-Polo M, Vyas S, Barmak BA, Galluci GO, Att W, Krishnamurthy VR. Artificial intelligence applications in implant dentistry: a systematic review. J Prosthet Dent. 2023;129:293–300. doi: 10.1016/j.prosdent.2021.05.008. [DOI] [PubMed] [Google Scholar]
- 13.Shetty S, Gali S, Augustine D, Sv S. Artificial intelligence systems in dental shade-matching: a systematic review. J Prosthodont. 2024;33:519–532. doi: 10.1111/jopr.13805. [DOI] [PubMed] [Google Scholar]
- 14.Bernauer SA, Zitzmann NU, Joda T. The use and performance of artificial intelligence in prosthodontics: a systematic review. Sensors (Basel) 2021;21:6628. doi: 10.3390/s21196628. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ali IE, Tanikawa C, Chikai M, Ino S, Sumita Y, Wakabayashi N. Applications and performance of artificial intelligence models in removable prosthodontics: a literature review. J Prosthodont Res. 2024;68:358–367. doi: 10.2186/jpr.JPR_D_23_00073. [DOI] [PubMed] [Google Scholar]
- 16.Chau RCW, Hsung RT, McGrath C, Pow EHN, Lam WYH. Accuracy of artificial intelligence-designed single-molar dental prostheses: a feasibility study. J Prosthet Dent. 2024;131:1111–1117. doi: 10.1016/j.prosdent.2022.12.004. [DOI] [PubMed] [Google Scholar]
- 17.Cho JH, Çakmak G, Yi Y, Yoon HI, Yilmaz B, Schimmel M. Tooth morphology, internal fit, occlusion and proximal contacts of dental crowns designed by deep learning-based dental software: a comparative study. J Dent. 2024;141:104830. doi: 10.1016/j.jdent.2023.104830. [DOI] [PubMed] [Google Scholar]
- 18.Yilmaz B, Marques VR, Donmez MB, Cuellar AR, Lu WE, Abou-Ayash S, Çakmak G. Influence of 3D analysis software on measured deviations of CAD-CAM resin crowns from virtual design file: an in-vitro study. J Dent. 2022;118:103933. doi: 10.1016/j.jdent.2021.103933. [DOI] [PubMed] [Google Scholar]
- 19.Nagata K, Muromachi K, Kouzai Y, Inaba K, Inoue E, Fuchigami K, Nihei T, Atsumi M, Kimoto K, Kawana H. Fit accuracy of resin crown on a dental model fabricated using fused deposition modeling 3D printing and a polylactic acid filament. J Prosthodont Res. 2023;67:144–149. doi: 10.2186/jpr.JPR_D_21_00325. [DOI] [PubMed] [Google Scholar]
- 20.Cho JH, Yi Y, Choi J, Ahn J, Yoon HI, Yilmaz B. Time efficiency, occlusal morphology, and internal fit of anatomic contour crowns designed by dental software powered by generative adversarial network: a comparative study. J Dent. 2023;138:104739. doi: 10.1016/j.jdent.2023.104739. [DOI] [PubMed] [Google Scholar]
- 21.Choi J, Ahn J, Park JM. Deep learning-based automated detection of the dental crown finish line: an accuracy study. J Prosthet Dent. 2024;132:1286.e1–1286.e9. doi: 10.1016/j.prosdent.2023.11.018. [DOI] [PubMed] [Google Scholar]
- 22.Rençber Kızılkaya A, Kara A. Impact of different CAD software programs on marginal and internal fit of provisional crowns: an in vitro study. Heliyon. 2024;10:e24205. doi: 10.1016/j.heliyon.2024.e24205. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Aktaş N, Bani M, Ocak M, Bankoğlu Güngör M. Effects of design software program and manufacturing method on the marginal and internal adaptation of esthetic crowns for primary teeth: a microcomputed tomography evaluation. J Prosthet Dent. 2024;131:519.e1–519.e9. doi: 10.1016/j.prosdent.2023.12.010. [DOI] [PubMed] [Google Scholar]
- 24.Daou EE, Ounsi H, Özcan M, Al-Haj Husain N, Salameh Z. Marginal and internal fit of pre-sintered Co-Cr and zirconia 3-unit fixed dental prostheses as measured using microcomputed tomography. J Prosthet Dent. 2018;120:409–414. doi: 10.1016/j.prosdent.2018.01.006. [DOI] [PubMed] [Google Scholar]
- 25.Ekici Z, Kılıçarslan MA, Bilecenoğlu B, Ocak M. Micro-CT Evaluation of the marginal and internal fit of crown and inlay restorations fabricated via different digital scanners belonging to the same CAD-CAM system. Int J Prosthodont. 2021;34:381–389. doi: 10.11607/ijp.6822. [DOI] [PubMed] [Google Scholar]
- 26.Rizonaki M, Jacquet W, Bottenberg P, Depla L, Boone M, De Coster PJ. Evaluation of marginal and internal fit of lithium disilicate CAD-CAM crowns with different finish lines by using a micro-CT technique. J Prosthet Dent. 2022;127:890–898. doi: 10.1016/j.prosdent.2020.11.027. [DOI] [PubMed] [Google Scholar]
- 27.Nagata K, Inaba K, Kimoto K, Kawana H. Accuracy of dental models fabricated using recycled poly-lactic acid. Materials (Basel) 2023;16:2620. doi: 10.3390/ma16072620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lin GSS, Ng YS, Ghani NRNA, Chua KH. Revolutionising dental technologies: a qualitative study on dental technicians’ perceptions of artificial intelligence integration. BMC Oral Health. 2023;23:690. doi: 10.1186/s12903-023-03389-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hiltunen K, Mäntylä P, Vehkalahti MM. Age- and time-related trends in oral health care for patients aged 60 years and older in 2007-2017 in Public Oral Health Services in Helsinki, Finland. Int Dent J. 2021;71:321–327. doi: 10.1016/j.identj.2020.12.006. [DOI] [PMC free article] [PubMed] [Google Scholar]