Evaluation of an artificial intelligence-based algorithm for automated localization of craniofacial landmarks.

OBJECTIVES: Due to advancing digitalisation, it is of interest to develop standardised and reproducible fully automated analysis methods of cranial structures in order to reduce the workload in diagnosis and treatment planning and to generate objectifiable data. The aim of this study was to train and evaluate an algorithm based on deep learning methods for fully automated detection of craniofacial landmarks in cone-beam computed tomography (CBCT) in terms of accuracy, speed, and reproducibility. MATERIALS AND METHODS: A total of 931 CBCTs were used to train the algorithm. To test the algorithm, 35 landmarks were located manually by three experts and automatically by the algorithm in 114 CBCTs. The time and distance between the measured values and the ground truth previously determined by an orthodontist were analyzed. Intraindividual variations in manual localization of landmarks were determined using 50 CBCTs analyzed twice. RESULTS: The results showed no statistically significant difference between the two measurement methods. Overall, with a mean error of 2.73 mm, the AI was 2.12% better and 95% faster than the experts. In the area of bilateral cranial structures, the AI was able to achieve better results than the experts on average. CONCLUSION: The achieved accuracy of automatic landmark detection was in a clinically acceptable range, is comparable in precision to manual landmark determination, and requires less time. CLINICAL RELEVANCE: Further enlargement of the database and continued development and optimization of the algorithm may lead to ubiquitous fully automated localization and analysis of CBCT datasets in future routine clinical practice.

Verification of a digital approach for three-dimensional evaluation of marginal and internal fit.

STATEMENT OF PROBLEM: Internal fit is an important aspect of indirect restorations, but methods for the 3-dimensional (3D) measurement of absolute marginal and intaglio fit are sparse. PURPOSE: The purpose of this in vitro study was to evaluate an innovative 3D measurement method (AIXFit) based on intraoral scanning data for analyzing the fit of dental restorations. MATERIAL AND METHODS: For the evaluation of AIXFit, 12 monolithic zirconia crowns were fabricated on typodont preparations. The fit was measured digitally with the AIXFit system and compared with the results obtained from an established 2-dimensional (2D) sectional procedure. To compare the values of both methods at identical locations, a common reference system was developed, with each die fixed in a gypsum stand with reference points. Using an intraoral scanner (True Definition), each die with its reference points and the intaglio surface of the finished crown were digitalized as standard tessellation language files. The AIXFit software program, with a specially developed best-fit algorithm, was used to match the intaglio surface of the crown with the surface of the preparation. The virtual cement gap was calculated over the entire surface and returned values for x≥0 µm. A 2D comparison method involved adhesively fixing the crown to the die and sectioning it into 4 parts with a diamond band saw. The thickness of the cement gap was determined under a light microscope at ×100 magnification at 5 defined measuring points per quarter, so that a total of 240 measurements were available for comparison. A software program (Blender Foundation) was used to superimpose the data from the AIXFit system with the data from the 2D method and to compare the cement gaps at the same locations. The agreement between these methods was verified using paired t tests and determine correlation coefficients (α=.05). RESULTS: The mean ±standard deviation difference between the AIXFit and 2D methods was 6.7 ±29 µm). Two 1-sided tests showed statistical equivalence between the methods of measurement when considering an interval between -20 and +20 µm. The correlation coefficients showed a positive association for both methods (r=.931). CONCLUSIONS: The AIXFit software program appeared to be accurate for the digital measurement of internal fit when using the True Definition scanner. It enabled a cast-free workflow and allowed the analysis of the entire intaglio surface.

Guided discovery learning: A follow-up study of try-it-yourself surgery and subsequent video-assisted teaching for oral surgical skills training.

INTRODUCTION: The purpose of this follow-up study was to assess the effects of discovery learning and subsequent standardised teaching on students' advanced dental surgical skills. MATERIALS AND METHODS: A total of 29 dental students who had no theoretical education on the Rehrmann flap were included in the skills training programme. Without prior teaching, the students were initially asked to perform a discovery surgery with a subsequent multiple-choice test (MCT) on the first day. This was followed by a video-assisted teaching lecture. On the same day and after 122 days, the surgery and MCT were repeated. Aside from the practical and theoretical assessment forms, the training was evaluated by the students themselves. RESULTS: Regarding the evaluation of surgery (EOS), significant improvements were measured between the first surgery and the second and third surgeries (p > .001). In addition, the theoretical test scores showed significant improvements after the first test when compared with the second (p = .004) and third tests (p < .001). For both assessments, no significant difference was found between the second and third attempts. The MCT and EOS in the second and third attempts strongly correlated (r = .77 and r = .71 respectively). CONCLUSION: The dental students in this study successfully learned a complex oral surgical procedure, the Rehrmann flap technique, for closing oro-antral communications. The participants indicated their high satisfaction with the teaching approach. After 122 days of follow-up, their practical and theoretical test scores remained high and presented no significant difference, which suggests that the newly learnt individual skills were retained.

Cemented versus screw-retained posterior implant-supported single crowns: A 24-month randomized controlled clinical trial.

OBJECTIVES: To compare the incidence of biological and technical complications of cemented and screw-retained monolithic lithium-disilicate implant-supported posterior single crowns. MATERIAL AND METHODS: Forty-one subjects with a total of 56 implants received randomly allocated 28 cemented and 28 screw-retained crowns. In the screw-retained group, monolithic lithium-disilicate restorations were luted to titanium bases extraorally. In the cemented group, monolithic lithium-disilicate crowns were cemented on individualized titanium abutments intraorally. All restorations were examined according to modified FDI criteria within 2 weeks of inserting the crowns (baseline) and after 12 (n = 46) and 24 (n = 43) months. Bone loss was evaluated by standardized radiographs at baseline and 12 months. RESULTS: After 12 months, the incidence of mucositis (positive bleeding on probing) was 14.2% (screw-retained) and 17.9% (cement-retained). The gingival and plaque index and a mean marginal bone loss between 0.03-0.15 mm showed no significant difference between the groups. In the cemented group, cement residues were detected at baseline at two restorations (6.9%) by radiographic examination. A complete digital workflow was realized in most cases (85.7%). At 24 months, no restoration had failed, and no chipping of the ceramic had occurred. In the screw-retained group, screw loosening occurred in one implant. In both groups, there was obvious deterioration in the quality of 32% of the occlusal and of 18% of the proximal contact points. CONCLUSIONS: The type of retention mode of monolithic implant-retained lithium-disilicate posterior crowns had no influence on the biological and technical complication rate.

Objective and Subjective Comparisons of Split Thickness Skin Graft and Full Thickness Skin Graft for Radial Forearm Flap Donor Sites Using a New Measuring Method.

ABSTRACT: A split thickness skin graft (STSG) or a full thickness skin graft (FTSG) are commonly used for donor site closure after raising a radial forearm flap. The aesthetic outcome of the donor site is frequently not satisfying for the patient. This study evaluated the aesthetic outcome of the donor site of a radial forearm flap covered with an STSG or FTSG using an objective measurement method. The forearms of 30 patients (15 FTSG, 15 STSG) were scanned with an optical three-dimensional scanner. The surface of the forearm at the skin transplant was cut out and filled by an automatic hole-filling algorithm. The mean surface deviation between the original and the filled forearm was acquired. This method was validated with a control group of 15 volunteers. A questionnaire investigating aesthetical and healing satisfaction, postoperative pain and scarring was filled in by the patients. The mean surface difference in the STSG group was 0.9907 ±â€Š0.3120 and 0.6177 ±â€Š0.2245 mm in the FTSG group. The difference in the surface deviation between STSG and FTSG groups was significant (P = 0.0009). Correlations between the surface deviation and the subjective aesthetical satisfaction of the patient were not significant. In the questionnaires no significant differences between STSG and FTSG group were measured. Both FTSGs and STSGs resulted in good aesthetical outcomes after closure of a radial forearm flap donor site. Scanning the radial forearm flap donor site with an optical three-dimensional scanner supplied an objective, fast, and reliable measuring method of aesthetical outcomes.

Do "cut out-rescan" procedures have an impact on the accuracy of intraoral digital scans?

STATEMENT OF PROBLEM: The software programs of digital intraoral scanners typically offers the option to cut out areas from 3D casts, to do rescans, and to merge them with the initial scan. However, evidence of whether this procedure has an impact on the accuracy of the scan is lacking. PURPOSE: The purpose of this study was to determine whether "cut out-rescan" procedures change the accuracy of a 3D cast. MATERIAL AND METHODS: A maxillary master cast was digitized with an industrial structured light scanner to obtain a digital reference cast. This master cast was repeatedly scanned by 3 intraoral scanners: TRIOS 3 [TR], Cerec Primescan [PR], and Cerec Omnicam [OM]. The scan data were duplicated, and the posterior area from the right lateral incisor was cut out and rescanned to obtain complete-arch casts containing the rescanned data [TR_rs], [PR_rs], and [OM_rs]. The trueness and precision of the scans were evaluated by superimposing procedures of the relevant data sets. To evaluate statistical differences, either the Mann-Whitney U test or the t test was used (α=.05). RESULTS: The median precision values of the complete-arch scan data was 19 µm for [OM] and [TR], whereas the median for [PR] was 14 µm. In the "cut out-rescanned" data group, the values were 25 µm for [OM_rs], 16 µm for [TR_rs], and 14 µm for [PR_rs]. Statistically significant differences were found among the scanners [OM]/[PR], [OM_rs]/[TR_rs], and [TR_rs]/[PR_rs]. The mean ± standard deviation values of trueness for the complete-arch scan data were 54 ±4 µm for [OM], 42 ±5 µm for [TR], and 30 ±2 µm for [PR]. In the group of the "cut out-rescanned" data, the mean trueness results were 55 ± 6 µm for [OM_rs], 38 ±5 µm for [TR_rs], and 31 ±5 µm for [PR_rs]. Significant differences were found among the complete-arch scan data and the "cut out-rescanned" data of the different scanners, but not between the complete-arch scan data and the "cut out-rescanned" data within one scanning system. CONCLUSIONS: Significant differences were found among the scanners, but "cut out-rescan" procedures did not affect the accuracy within each scanning system.

Objective computerised assessment of residual ridge resorption in the human maxilla and maxillary sinus pneumatisation.

OBJECTIVES: Atrophic resorption of the maxillary alveolar ridge is a complication that makes implantological rehabilitation critical. Our aim was to develop a novel computer aided procedure for the accurate quantitative assessment of maxillary residual ridge resorption including pneumatisation of the maxillary sinus that goes beyond previously described approaches and to apply it to a large dataset. MATERIALS AND METHODS: To develop and refine the method, we performed a retrospective analysis using computed tomography data from 405 patients to generate segmented, three-dimensional models of zygomaticomaxillary bones and maxillary sinuses. Using anatomical landmarks and orientation lines or planes, all models were aligned automatically to subsequently generate cross-sectional images (n = 2835), enabling the classification of atrophy as well as the quantification of volumes and caudal extensions of the maxillary sinuses. RESULTS: We developed and implemented an accurate and reproducible workflow for the semi-automated analysis of volumetric maxillary images. Comprehensive statistical analysis of the large quantitative dataset revealed various correlations of maxillary process heights and sinus volumes with atrophy class, age and region and identified conjectural trends over the patient group. CONCLUSIONS: The method was used successfully to process a large dataset to classify atrophy, to measure alveolar height parameters, and to quantify maxillary sinus volume, bottom volume and pneumatisation. CLINICAL RELEVANCE: Apart from the anthropometric value of the generated dataset, the method could be applied to provide additional and more accurate data to assess the necessity of bone augmentation in the context of three-dimensional planning before implantation.

Stress distribution in mandibular donor site after harvesting bone grafts of various sizes from the ascending ramus of a dentate mandible by finite element analysis.

PURPOSE: Harvesting bone from the ascending ramus of the mandible is a common procedure. However, mandibular fracture may occur after grafting bone blocks. This study aimed to investigate the resulting force distribution of stress and strain in the mandibular donor site after harvesting bone grafts of different sizes and various loadings. METHODS: Finite element analysis was performed for virtual harvesting of bone blocks of nine different sizes between 15 × 20 and 25 × 30 mm and three different chewing loads (incisal, ipsilateral and contralateral). von Mises stress and first principal stress distributions were measured. RESULTS: von Mises stress was distributed between 35.01 (10 × 15 mm graft, incisal load) and 333.25 MPa (30 × 20 mm graft ipsilateral load), whereas first principal stress distributions were between 48.27 (10 × 15 mm graft, incisal load) and 414.69 MPa (30 × 20 mm graft ipsilateral load). In general, the least stress was observed with incisal load followed by ipsilateral load and finally contralateral load. The critical value of 133 MPa was found after removing almost all grafts with a width of 20 or 30 mm. CONCLUSIONS: Incisal loading led to less stress compared with contralateral and ipsilateral loads. Increasing graft size led to increasing weakness of the donor site. Graft width exerted a greater influence on stress development than its height. CLINICAL RELEVANCE: Ipsilateral chewing and increasing width of the bone graft result in maximum stress in the mandibular donor side, and critical values regarding to the possibility of fractures are already to expect from a graft size of 20 × 15 mm.

Functional MR Imaging Mapping of Human Articular Cartilage Response to Loading.

Purpose To determine if multiparametric magnetic resonance (MR) imaging mapping can be used to quantify the response to loading of histologically intact human knee cartilage. Materials and Methods Institutional review board approval and written informed consent were obtained. Twenty macroscopically intact cartilage-bone samples were obtained from the central lateral femoral condyles in 11 patients undergoing total knee replacement. A clinical 3.0-T MR imaging system was used to generate T1, T1ρ, T2, and T2* maps with inversion recovery, spin-lock multiple gradient-echo, multiple spin-echo, and multiple gradient-echo sequences. Serial mapping was performed at three defined strain levels (strain 0 [δ0], 0%; strain 1 [δ1/2], 19.8% ± 4.6 [standard deviation]; strain 2 [δ1], 39.5% ± 9.3) by using displacement-controlled static indentation loading. The entire sample and specific cartilage zones (superficial zone [SZ], transitional zone [TZ], and deep zone [DZ]) and regions (subpistonal area [SPA] and peripistonal area [PPA]) were defined as regions of interest. Upon log transformation, repeated measures analysis of variance was used to detect groupwise regional and zonal differences. Load-induced relative changes were determined and analyzed by using paired Student t test and Spearman correlation. Biomechanical testing (unconfined compression) and histologic assessment (Mankin score) served as the reference standard. Results All samples were histologically intact. Strain-related decreases were found at the SZ and TZ for T1 and T2*; for T1ρ, increases were seen in all zones; and for T2, increases were seen at the SZ and PPA only. Significant parameter changes in the entire sample depth of SPA versus PPA were found for δ1/2 (T1ρ, 14% ± 12 vs 6% ± 9) and δ1 (T1, -4% ± 5 vs -1% ± 3; T1ρ, 13% ± 12 vs 7% ± 7; T2*, -9% ± 12 vs -2% ± 8). SPA versus PPA changes were significant at the SZ and TZ (T1), TZ and DZ (T1ρ), and SZ (T2*). No significant correlations were found between relative changes and biomechanical or histologic parameters. Conclusion Serial multiparametric MR imaging mapping can be used to evaluate cartilage beyond mere static analysis and may provide the basis for more refined graduation strategies of cartilage degeneration. © RSNA, 2016 Online supplemental material is available for this article.

Locking versus nonlocking plates in mandibular reconstruction with fibular graft--a biomechanical ex vivo study.

OBJECTIVES: The main goal of the present study was to compare the biomechanical stability of locking plates and conventional miniplate combinations in human mandibles reconstructed with fibular grafts. MATERIALS AND METHODS: A specially developed and well-proven testing device reproduced the in vivo loading conditions on the mandible. Cadaveric human mandibles (n = 12) reconstructed with harvested human fibular bone grafts were divided into two groups, and different osteosynthesis systems were applied using two lines of plates per osteotomy. On the test apparatus, the specimens were stressed to failure, and interfragmentary movement was monitored and quantified with a contact-free optical measurement system. RESULTS: The relevant interfragmentary movement results from a Euclidean summary calculation which considered all three spatial angles around the axes. Using values up to a maximum load of 300 N, the conventional six-hole miniplates (profile 1.0) had an average value of 7.45° ± 1.46°, and the locking six-hole plates (profile 1.3) had an average value of 12.16° ± 2.37° for rotational interfragmentary movement. The miniplate system exhibited a significantly superior performance in fixation compared to the fixed-angle system (p < 0.05). CONCLUSION: According to these biomechanical experiments, both osteosynthesis devices provided sufficient stabilization at loads of up to 300 N. The six-hole miniplate system provided better stabilization of the osteotomy gap for mandibles reconstructed with fibular grafts. CLINICAL RELEVANCE: The osteosynthesis system is essential for primary stability and the avoidance of pseudarthrosis formation. This study demonstrates that the miniplates provide sufficient stabilization and offers a method to improve fixation in reconstructed mandibles.

High-fidelity human patient simulators compared with human actors in an unannounced mass-casualty exercise.

High-fidelity simulators (HFSs) have been shown to prompt critical actions at a level equal to that of trained human actors (HAs) and increase perceived realism in intrahospital mass-casualty incident (MCI) exercises. For unannounced prehospital MCI exercises, however, no data are available about the feasibility of incorporating HFSs. This case report describes the integration of HFSs in such an unannounced prehospital MCI drill with HAs and provides data about the differences concerning triage, treatment, and transport of HFSs and HAs with identical injury patterns. For this purpose, 75 actors and four high-fidelity simulators were subdivided into nine groups defined by a specific injury pattern. Four HFSs and six HAs comprised a group suffering from traumatic brain injury and blunt abdominal trauma. Triage results, times for transport, and number of diagnostic and therapeutic tasks were recorded. Means were compared by t test or one-way ANOVA. Triage times and results did not differ between actors and simulators. The number of diagnostic (1.25, SD = 0.5 in simulators vs 3.5, SD = 1.05 in HAs; P = .010) and therapeutic tasks (2.0, SD = 1.6 in simulators vs 4.8, SD = 0.4 in HAs; P = .019) were significantly lower in simulators. Due to difficulties in treating and evacuating the casualties from the site of the accident in a timely manner, all simulators died. Possible causal factors and strategies are discussed, with the aim of increasing the utility of simulators in emergency medicine training.

Effects of differing thickness and mechanical properties of cement on the stress levels and distributions in a three-unit zirconia fixed prosthesis by FEA.

PURPOSE: This study analyzed the impact of cement layer thickness (CLT) and Young's modulus of the cement on the stress distribution in a three-unit zirconia fixed dental prosthesis (FDP) and in the bonding interfaces by means of finite element method. MATERIALS AND METHODS: A 3D finite element model was created from a stylized three-unit FDP-cement-tooth/socket system. The pulp and the periodontal ligament were not modeled. Two CLTs (50 and 150 µm) and two values of Young's modulus of the cement (4.9 for simulation of resin cement, 20.1 GPa for glass ionomer cement) were evaluated. A 500 N static vertical load was applied at the central fossa of the pontic to calculate maximum displacement in the framework and maximum principal stresses in both framework and bonding interfaces. RESULTS: The simulated results showed that the Young's modulus affected stress occurrence only in the cement interface. Lower moduli were associated with less stress. The thickness of the cement layer influenced the maximum principal stress in both the FDP and in the cement layer itself. Thicker cement layers led to higher stresses in the framework but lower stresses in the cement layer. Maximum displacement was less dependent of the investigated variables. During all trials, the location of the maximum principal stress did not change. Maximum stress concentrations were observed at the lower embrasures of the connector areas and in the bonding layer at the cervical margin of the preparation. CONCLUSIONS: Choosing cements with a preferably low Young's modulus in combination with a CLT as small as possible might increase the clinical survival rate.

Optimal untwisting of the orbital bandeau in unicoronal craniosynostosis correction: A finite element analysis.

BACKGROUND: Surgical correction of unicoronal craniosynostosis (UCS) is highly complex due to its asymmetric appearance. Although fronto-orbital advancement (FOA) is a versatile technique for craniosynostosis correction, harmonization of the orbital bandeau in UCS is difficult to predict. This study evaluates the biomechanics of the orbital bandeau using different patterns and varying characteristics of inner cortical bone layer osteotomies in a finite element (FE) analysis. METHOD: An FE model was created using the computed tomography (CT) scan of a 6.5-month-old male infant with a right-sided UCS. The unaffected side of the orbital bandeau was virtually mirrored, and anatomical correction of the orbital bandeau was simulated. Different combinations of osteotomy patterns, numbers, depths, and widths were examined (n = 48) and compared to an uncut model. RESULTS: Reaction forces and maximum stress values differed significantly (p < 0.01) among osteotomy patterns and between each osteotomy characteristic. Regardless of the osteotomy pattern, higher numbers of osteotomies significantly (p < 0.05) correlated with reductions in reaction force and maximum stress. An X-shaped configuration with three osteotomies deep and wide to the bone was biomechanically the most favorable model. CONCLUSION: Inner cortical bone layer osteotomy might be an effective modification to the conventional FOA approach in terms of predictable shaping of the orbital bandeau.

Accuracy of the surgical execution of virtually planned deep circumflex iliac artery flaps and their appropriateness for masticatory rehabilitation.

BACKGROUND: Tumorous diseases of the jaw demand effective treatments, often involving continuity resection of the jaw. Reconstruction via microvascular bone flaps, like deep circumflex iliac artery flaps (DCIA), is standard. Computer aided planning (CAD) enhances accuracy in reconstruction using patient-specific CT images to create three-dimensional (3D) models. Data on the accuracy of CAD-planned DCIA flaps is scarce. Moreover, the data on accuracy should be combined with data on the exact positioning of the implants for well-fitting dental prosthetics. This study focuses on CAD-planned DCIA flaps accuracy and proper positioning for prosthetic rehabilitation. METHODS: Patients post-mandible resection with CAD-planned DCIA flap reconstruction were evaluated. Postoperative radiograph-derived 3D models were aligned with 3D models from the CAD plans for osteotomy position, angle, and flap volume comparison. To evaluate the DCIA flap's suitability for prosthetic dental rehabilitation, a plane was created in the support zone and crestal in the middle of the DCIA flap. The lower jaw was rotated to close the mouth and the distance between the two planes was measured. RESULTS: 20 patients (12 males, 8 females) were included. Mean defect size was 73.28 ± 4.87 mm; 11 L defects, 9 LC defects. Planned vs. actual DCIA transplant volume difference was 3.814 ± 3.856 cm³ (p = 0.2223). The deviation from the planned angle was significantly larger at the dorsal osteotomy than at the ventral (p = 0.035). Linear differences between the planned DCIA transplant and the actual DCIA transplant were 1.294 ± 1.197 mm for the ventral osteotomy and 2.680 ± 3.449 mm for the dorsal (p = 0.1078). The difference between the dental axis and the middle of the DCIA transplant ranged from 0.2 mm to 14.8 mm. The mean lateral difference was 2.695 ± 3.667 mm in the region of the first premolar. CONCLUSION: The CAD-planned DCIA flap is a solution for reconstructing the mandible. CAD planning results in an accurate reconstruction enabling dental implant placement and dental prosthetics.

Development of a universal cutting guide for raising deep circumflex iliac artery flaps.

PURPOSE: The deep circumflex iliac crest flap (DCIA) is used for the reconstruction of the jaw. For fitting of the transplant by computer-aided planning (CAD), a computerized tomography (CT) of the jaw and the pelvis is necessary. Ready-made cutting guides save a pelvic CT and healthcare resources while maintaining the advantages of the CAD planning. METHODS: A total of 2000 CTs of the pelvis were divided into groups of 500 by sex and age (≤ 45 and > 45 years). Three-dimensional (3D) pelvis models were aligned and averaged. Cutting guides were designed on the averaged pelvis for each group and an overall averaged pelvis. The cutting guides and 50 randomly selected iliac crests (10 from each group and 10 from the whole collective) were 3D printed. The appropriate cutting guide was mounted to the iliac crest and a cone beam CT was performed. The thickness of the space between the iliac crest and the cutting guide was evaluated. RESULTS: Overall the mean thickness of the space was 2.137 mm and the mean volume of the space was 4513 mm3. The measured values were significantly different between the different groups. The overall averaged group had not the greatest volume, maximum thickness and mean thickness of the space. CONCLUSION: Ready-made cutting guides for the DCIA flap fit to the iliac crest and make quick and accurate flap raising possible while radiation dose and resources can be saved. The cutting guides fit sufficient to the iliac crest and should keep the advantages of a standard CAD planning.

The impact of orbital floor defect ratio on changes in the inferior rectus muscle and prediction of posttraumatic enophthalmos - A cadaver study.

BACKGROUND: Orbital floor fractures result in critical changes in the shape and inferior rectus muscle (IRM) position. Radiological imaging of IRM changes can be used for surgical decision making or prediction of ocular symptoms. Studies with a systematic consideration of the orbital floor defect ratio in this context are missing in the literature. Accordingly, this study on human cadavers aimed to systematically investigate the impact of the orbital floor defect ratio on changes in the IRM and the prediction of posttraumatic enophthalmos. METHODS: Seventy-two orbital floor defects were placed in cadaver specimens using piezosurgical removal. The orbital defect area (ODA), orbital floor area (OFA), position and IRM shape, and enophthalmos were measured using computed tomography (CT) scans. RESULTS: The ODA/OFA ratio correlated significantly (p < 0.001) with the shape (Spearman's rho: 0.558) and position (Spearman's rho: 0.511) of the IRM, and with enophthalmos (Spearman's rho: 0.673). Increases in the ODA/OFA ratio significantly rounded the shape of the IRM (ß: 0.667; p < 0.001) and made a lower position of the IRM more likely (OR: 1.093; p = 0.003). In addition, increases in the ODA/OFA ratio were significantly associated with the development of relevant enophthalmos (OR: 1.159; p = 0.008), adjusted for the defect localization and shape of the IRM. According to receiver operating characteristics analysis (AUC: 0.876; p < 0.001), a threshold of ODA/OFA ratio ≥ 32.691 for prediction of the risk of development of enophthalmos yielded a sensitivity of 0.809 and a specificity of 0.842. CONCLUSION: The ODA/OFA ratio is a relevant parameter in the radiological evaluation of orbital floor fractures, as it increases the risk of relevant enophthalmos, regardless of fracture localization and shape of the IRM. Therefore, changes in the shape and position of the IRM should be considered in surgical treatment planning. A better understanding of the correlates of isolated orbital floor fractures may help to develop diagnostic scores and standardize therapeutic algorithms in the future.

Comparative assessment of 3D surface scanning systems in breast plastic and reconstructive surgery.

In this work, we compared accuracy, repeatability, and usability in breast surface imaging of 2 commercial surface scanning systems and a hand-held laser surface scanner prototype coupled with a patient's motion acquisition and compensation methodology. The accuracy of the scanners was assessed on an anthropomorphic phantom, and to evaluate the usability of the scanners on humans, thorax surface images of 3 volunteers were acquired. Both the intrascanner repeatability and the interscanner comparative accuracy were assessed. The results showed surface-to-surface distance errors inferior to 1 mm and to 2 mm, respectively, for the 2 commercial scanners and for the prototypical one. Moreover, comparable performances of the 3 scanners were found when used for acquiring the breast surface. On the whole, this study demonstrated that handheld laser surface scanners coupled with subject motion compensation methods lend themselves as competitive technologies for human body surface modeling.

Accuracy of four recent intraoral scanners with respect to two different ceramic surfaces.

OBJECTIVES: To investigate the complete arch accuracy of intraoral scanners (IOS) on two different ceramic surfaces. METHODS: Two maxillary master cast samples were prepared. The bases of both the master casts were made from zirconium oxide. The difference between the two casts was that the teeth of the [ZR] cast were produced from zirconium oxide and that of the [LD] cast were made of lithium disilicate glass-ceramic. Unlike the zirconia teeth of the [ZR] cast, the lithium disilicate teeth of the [LD] cast were glazed. The two master casts were digitized using a high-resolution scanner (Atos Compact Scan 5 M, GOM GmbH, Braunschweig, Germany) to obtain digital reference casts. Subsequently, each master cast was scanned 15 times using four IOSs. The IOSs were the Cerec Omnicam [OM], Primescan [PR], Trios 4 [TR4], and VivaScan [VS]. On surface comparison, the absolute mean deviation values were obtained for trueness and precision. For multiple comparisons, statistically significant differences were analyzed using one-way ANOVA and the Kruskal-Wallis H test. The p-value was adjusted to control for the increased risk of type I error (p < 0.0083). To compare the two means, the t-test and Mann-Whitney U test were used (p < 0.05). RESULTS: Trueness values for [ZR] ranged from 24.6 (±6.3) µm for [PR] and 77.1 (±8.3) µm for [OM]. Trueness values for [LD] were between 28.3 (±6.3) µm for [PR] and 72.8 (±15.6) µm for [OM]. Precision values for [ZR] ranged from 17.6 (±3.7) µm for [PR] to 37.3 (±9.9) µm for [OM]. Precision values for [LD] ranged from 17.5 (±3.6) µm for [PR] to 41.8 (±8.7) µm for [OM]. Statistically significant differences were found among all the IOSs (p < 0.0083). The trueness values of the four IOSs did not differ significantly (p < 0.05) with respect to either the [ZR] or [LD] cast. The precision values of [OM] and [VS] differed significantly with respect to the scanned surface. CONCLUSIONS: Complete arch scans achieved with the four IOSs showed significantly different trueness and precision results. [VS] and [OM] were more sensitive in terms of the scanned material. CLINICAL SIGNIFICANCE: The latest IOSs showed the required accuracy for complete arch digital impressions in-vitro investigations. These findings should be implemented under conditions relevant to complete arch deviations, such as the construction of occlusal splints, analysis of occlusal relationships, and long-span restorations. Clinicians should be aware that the clinically acceptable threshold varies depending on the purpose of the IOS.

Mandible segmentation from CT data for virtual surgical planning using an augmented two-stepped convolutional neural network.

PURPOSE: For computer-aided planning of facial bony surgery, the creation of high-resolution 3D-models of the bones by segmenting volume imaging data is a labor-intensive step, especially as metal dental inlays or implants cause severe artifacts that reduce the quality of the computer-tomographic imaging data. This study provides a method to segment accurate, artifact-free 3D surface models of mandibles from CT data using convolutional neural networks. METHODS: The presented approach cascades two independently trained 3D-U-Nets to perform accurate segmentations of the mandible bone from full resolution CT images. The networks are trained in different settings using three different loss functions and a data augmentation pipeline. Training and evaluation datasets consist of manually segmented CT images from 307 dentate and edentulous individuals, partly with heavy imaging artifacts. The accuracy of the models is measured using overlap-based, surface-based and anatomical-curvature-based metrics. RESULTS: Our approach produces high-resolution segmentations of the mandibles, coping with severe imaging artifacts in the CT imaging data. The use of the two-stepped approach yields highly significant improvements to the prediction accuracies. The best models achieve a Dice coefficient of 94.824% and an average surface distance of 0.31 mm on our test dataset. CONCLUSION: The use of two cascaded U-Net allows high-resolution predictions for small regions of interest in the imaging data. The proposed method is fast and allows a user-independent image segmentation, producing objective and repeatable results that can be used in automated surgical planning procedures.

In-vitro accuracy of casts for orthodontic purposes obtained by a conventional and by a printer workflow.

This in-vitro study was designed to investigate whether conventionally produced casts and printed casts for orthodontic purposes show comparable full-arch accuracy. To produce casts, either a conventional impression or a digital data set is needed. A fully dentate all ceramic master cast was digitized with an industrial scanner to obtain a digital reference cast [REF]. Intraoral scans [IOS] and alginate impressions were taken from the master cast so that ten printed and ten gypsum casts were obtained. The printed casts [DLP] were digitized by an industrial scanner and as well as the gypsum casts [GYPSUM]. The following absolute mean trueness evaluations by superimposition were accomplished: [REF vs. GYPSUM]; [REF vs. DLP]; [REF vs. IOS]; [IOS vs. DLP]. For precision analysis the data sets of [GYPSUM], [IOS] and [DLP] were available. The absolute mean trueness values were 68 µm ± 15 µm for [REF vs. GYPSUM], 46 µm ± 4 µm for [REF vs. DLP], 20 µm ± 2 µm for [REF vs. IOS] and 41 µm ± 4 µm for [IOS vs. DLP]. [REF vs. GYPSUM] and [REF vs. DLP], [REF vs. IOS], [REF vs. DLP] and [IOS vs. DLP] showed statistically significant differences. The precision values were 56 µm ± 17 µm for [GYPSUM], 25 µm ± 9 µm for [DLP] and 12 µm ± 2 µm for [IOS] and differed significantly among each other. In the present study the print workflow revealed superior results in comparison to the conventional workflow. Due to contrary deviations in the [REF vs. IOS] and the [IOS vs. DLP] data sets the overall trueness deviations was enhanced.