Preinterventional Third-Molar Assessment Using Robust Machine Learning.

Machine learning (ML) models, especially deep neural networks, are increasingly being used for the analysis of medical images and as a supporting tool for clinical decision-making. In this study, we propose an artificial intelligence system to facilitate dental decision-making for the removal of mandibular third molars (M3M) based on 2-dimensional orthopantograms and the risk assessment of such a procedure. A total of 4,516 panoramic radiographic images collected at the Center of Dental Medicine at the University of Zurich, Switzerland, were used for training the ML model. After image preparation and preprocessing, a spatially dependent U-Net was employed to detect and retrieve the region of the M3M and inferior alveolar nerve (IAN). Image patches identified to contain a M3M were automatically processed by a deep neural network for the classification of M3M superimposition over the IAN (task 1) and M3M root development (task 2). A control evaluation set of 120 images, collected from a different data source than the training data and labeled by 5 dental practitioners, was leveraged to reliably evaluate model performance. By 10-fold cross-validation, we achieved accuracy values of 0.94 and 0.93 for the M3M-IAN superimposition task and the M3M root development task, respectively, and accuracies of 0.9 and 0.87 when evaluated on the control data set, using a ResNet-101 trained in a semisupervised fashion. Matthew's correlation coefficient values of 0.82 and 0.75 for task 1 and task 2, evaluated on the control data set, indicate robust generalization of our model. Depending on the different label combinations of task 1 and task 2, we propose a diagnostic table that suggests whether additional imaging via 3-dimensional cone beam tomography is advisable. Ultimately, computer-aided decision-making tools benefit clinical practice by enabling efficient and risk-reduced decision-making and by supporting less experienced practitioners before the surgical removal of the M3M.

Computer-assisted bone augmentation, implant planning and placement: An in vitro investigation.

AIM: To assess in vitro the workflow for alveolar ridge augmentation with customised 3D printed block grafts and simultaneous computer-assisted implant planning and placement. METHODS: Twenty resin mandible models with an edentulous area and horizontal ridge defect in the region 34-36 were scanned with cone beam computed tomography (CBCT). A block graft for horizontal ridge augmentation in the region 34-36 and an implant in the position 35 were digitally planned. Twenty block grafts were 3D printed out of resin and one template for guided implant placement were stereolithographically produced. The resin block grafts were positioned onto the ridge defects and stabilised with two fixation screws each. Subsequently, one implant was inserted in the position 35 through the corresponding template for guided implant placement. Optical scans of the study models together with the fixated block graft were performed prior to and after implant placement. The scans taken after block grafting were superimposed with the virtual block grafting plan through a best-fit algorithm, and the linear deviation between the planned and the achieved block positions was calculated. The precision of the block fixation was obtained by superimposing the 20 scans taken after grafting and calculating the deviation between the corresponding resin blocks. The superimposition between the scans taken after and prior to implant placement was performed to measure a possible displacement in the block position induced by guided implant placement. The (98-2%)/2 percentile value was determined as a parameter for surface deviation. RESULTS: The mean deviation in the position of the block graft compared to the virtual plan amounted to 0.79 ± 0.13 mm. The mean deviation between the positions of the 20 block grafts measured 0.47 ± 0.2 mm, indicating a clinically acceptable precision. Guided implant placement induced a mean shift of 0.16 ± 0.06 mm in the position of the block graft. CONCLUSIONS: Within the limitations of this in vitro study, it can be concluded that customised block grafts fabricated through CBCT, computer-assisted design and 3D printing allow alveolar ridge augmentation with clinically acceptable predictability and reproducibility. Computer-assisted implant planning and placement can be performed simultaneously with computer-assisted block grafting leading to clinically non-relevant dislocation of block grafts.

Systemic sclerostin antibody treatment increases osseointegration and biomechanical competence of zoledronic-acid-coated dental implants in a rat osteoporosis model.

Osseointegration of dental implants can be promoted by implant-surface modifications using bisphosphonate coatings. In addition, it is of clinical interest to promote peri-implant bone formation and to restore bony structure in low bone-mass patients. The present study evaluated a combination of an anti-resorptive zoledronic acid (ZOL) implant-coating and a systemically applied sclerostin antibody, a known bone anabolic treatment principle, versus sole sclerostin antibody treatment or ZOL implant-coating in a rat osteoporosis model. Uncoated reference surface implants or ZOL-coated implants (n = 64/group) were inserted into the proximal tibia of aged osteoporotic rats three months following ovariectomy. 32 animals of each group received once weekly sclerostin antibody therapy. Osseointegration was assessed 2 or 4 weeks post-implantation by ex vivo µCT, histology and biomechanical testing. Overall implant survival rate was 97 %. Histomorphology revealed pronounced bone formation along the entire implant length of ZOL-coated implants. At 4 weeks following implant insertion, bone-implant contact, cancellous bone mineral density and bone volume/tissue volume were significantly increased for the combination of ZOL and sclerostin antibody as compared to sclerostin antibody or ZOL implant-coating alone. Removal torque was also significantly increased in the combination therapy group relative to animals receiving only sclerostin antibody therapy or ZOL-coated implants. In an osteoporotic rat model, the combination of anti-resorptive ZOL implant-coating and systemically applied sclerostin antibody led to significantly increased peri-implant bone formation. Therefore, the combination of ZOL and the osteoanabolic sclerostin antibody was more effective than either agent alone.

Anatomical characteristics of maxillary sinus septa visualized by cone beam computed tomography.

Proper implant positioning in the posterior region of the edentulous maxilla commonly requires sinus floor elevation. Maxillary sinus septa increase the risk of membrane perforation during sinus floor elevation. The purpose of this retrospective, cone beam computed tomography (CBCT)-based study was to examine the frequency, number, location, and orientation of antral septa in the maxillary sinus. Further, possible associated factors were assessed. Measurements were performed on CBCT scans of 301 patients (602 sinuses). The data were analysed statistically with respect to patient age, sex, and dentition type. One or more septa were detected in 117 patients (38.9%). A total of 188 septa were found in the 602 sinuses (31.2%). Septa were most often coronally oriented (53.2%), followed by sagittal (24.5%) and transverse (22.3%) orientations. Septa were most often found in the region of the first and second molar (37.2%), followed by the posterior region of the third molar (33.0%) and the anterior region of the premolars and canines (29.8%). A significant association was found between edentulism and the presence of septa. For edentulous patients, the septa were most often transversally oriented. Maxillary sinus septa are encountered in every third patient. This may have an influence on the performance of sinus floor elevation.

Evaluation of oral prophylaxis during and after intensity-modulated radiotherapy due to head and neck cancer--a retrospective study.

OBJECTIVES: The aim of this study was to retrospectively analyze the influence of a prophylaxis protocol of head and neck tumor (HNT) patients during and after intensity-modulated radiotherapy (IMRT). MATERIAL AND METHOD: In this 5-year retrospective study (2009-2013), 70 (m 55, f 15; age range 28-8 years; median 58.7 years) out of 248 HNT patients of the Clinic of Cranio-Maxillofacial and Oral Surgery at the University of Zurich, Switzerland, fulfilled the inclusion criteria. Parameters of investigation were the salivary flow rates, possible dental foci and the dental status, oral side effects of radiotherapy, the prophylaxis protocol, and patient's compliance to this protocol. The following time points before during and after IMRT (6 weeks) were analyzed: prior to IMRT, 2-4 weeks, 6 weeks and 3, 6, and 12 months after the onset of radiotherapy. RESULT: Unstimulated salivary flow rate, pH value of unstimulated salivary, and stimulated salivary flow rate showed a significant reduction over time (p < 0.001). One year after IMRT, both unstimulated and stimulated salivary flow showed a statistically significant lower salivary flow. The number of caries-affected sites per patient was significantly higher for patients with low compliance to the prophylaxis protocol (mean: low compliance 1.36, high compliance 0.26). Almost 75% of the evaluated patients suffered immediate gustatory change, and 47.1% showed signs of radiostomatitis through IMRT. CONCLUSIONS: High compliance to the prophylaxis protocol during and after radiotherapy is a key factor for the reduction of radiation side effects on dental hard tissue. CLINICAL RELEVANCE: High compliance to a monitored prophylaxis program is crucial for patients after head and neck surgery.

Oral lichen planus (OLP), oral lichenoid lesions (OLL), oral dysplasia, and oral cancer: retrospective analysis of clinicopathological data from 2002-2011.

INTRODUCTION: This 10-year retrospective study analyzed the incidence of malignant transformation of oral lichen planus (OLP). The study also included dysplasia and oral lichenoid lesion (OLL) in the initial biopsy as a potential differential diagnosis. MATERIAL AND METHODS: A total of 692 scalpel biopsies were taken from 542 patients (207 [38.2%] men and 335 [61.8%] women). Clinical and histopathological parameters were analyzed. RESULTS: The parameters gender (p = 0.022) and smoking behavior (p < 0.001) were significantly associated with the severity of diagnosis. Mucosal lesions with an ulcerative appearance (p = 0.006) and those located on the floor of the mouth (p < 0.001) showed significantly higher degrees of dysplasia or were diagnosed as oral squamous cell carcinoma (OSCC). Smoking and joint disease appeared to be significant risk factors. Treatment with tretinoin in different concentrations (0.005-0.02%) significantly improved diagnosis. Twelve patients (8 female, 4 male) showed malignant transformation to OSCC within an average period of 1.58 years. The malignant transformation rate (MTR) was higher for OLL (4.4%) than OLP (1.2%). If the first biopsy showed intraepithelial neoplasia, the risk of developing OSCC increased (by 3.5% for squamous intraepithelial neoplasia (SIN) II and by 6.7% for SIN III). CONCLUSION: Although we cannot rule out that OLP is a premalignant oral condition, we can confirm that OLP had the lowest MTR of all diagnoses.

Chondroitin sulfate and sulfated hyaluronan-containing collagen coatings of titanium implants influence peri-implant bone formation in a minipig model.

An improved osseous integration of dental implants in patients with lower bone quality is of particular interest. The aim of this study was to evaluate the effect of artificial extracellular matrix implant coatings on early bone formation. The coatings contained collagen (coll) in conjunction with either chondroitin sulfate (CS) or sulfated hyaluronan (sHya). Thirty-six screw-type, grit-blasted, and acid-etched titanium implants were inserted in the mandible of 6 minipigs. Three surface states were tested: (1) uncoated control (2) coll/CS (3) coll/sHya. After healing periods of 4 and 8 weeks, bone implant contact (BIC), bone volume density (BVD) as well as osteoid related parameters were measured. After 4 weeks, control implants showed a BIC of 44% which was comparable to coll/CS coated implants (48%) and significantly higher compared to coll/sHya coatings (37%, p = 0.012). This difference leveled out after 8 weeks. No significant differences could be detected for BVD values after 4 weeks and all surfaces showed reduced BVD values after 8 weeks. However, at that time, BVD around both, coll/CS (30%, p = 0.029), and coll/sHya (32%, p = 0.015), coatings was significantly higher compared to controls (22%). The osteoid implant contact (OIC) showed no significant differences after 4 weeks. After 8 weeks OIC for controls was comparable to coll/CS, the latter being significantly higher compared to coll/sHya (0.9% vs. 0.4%, p = 0.012). There were no significant differences in osteoid volume density. In summary, implant surface coatings by the chosen organic components of the extracellular matrix showed a certain potential to influence osseointegration in vivo.

Osseointegration of biochemically modified implants in an osteoporosis rodent model.

The present study examined the impact of implant surface modifications on osseointegration in an osteoporotic rodent model. Sandblasted, acid-etched titanium implants were either used directly (control) or were further modified by surface conditioning with NaOH or by coating with one of the following active agents: collagen/chondroitin sulphate, simvastatin, or zoledronic acid. Control and modified implants were inserted into the proximal tibia of aged ovariectomised (OVX) osteoporotic rats (n = 32/group). In addition, aged oestrogen competent animals received either control or NaOH conditioned implants. Animals were sacrificed 2 and 4 weeks post-implantation. The excised tibiae were utilised for biomechanical and morphometric readouts (n = 8/group/readout). Biomechanical testing revealed at both time points dramatically reduced osseointegration in the tibia of oestrogen deprived osteoporotic animals compared to intact controls irrespective of NaOH exposure. Consistently, histomorphometric and microCT analyses demonstrated diminished bone-implant contact (BIC), peri-implant bone area (BA), bone volume/tissue volume (BV/TV) and bone-mineral density (BMD) in OVX animals. Surface coating with collagen/chondroitin sulphate had no detectable impact on osseointegration. Interestingly, statin coating resulted in a transient increase in BIC 2 weeks post-implantation; which, however, did not correspond to improvement of biomechanical readouts. Local exposure to zoledronic acid increased BIC, BA, BV/TV and BMD at 4 weeks. Yet this translated only into a non-significant improvement of biomechanical properties. In conclusion, this study presents a rodent model mimicking severely osteoporotic bone. Contrary to the other bioactive agents, locally released zoledronic acid had a positive impact on osseointegration albeit to a lesser extent than reported in less challenging models.

Comparison of zirconia and titanium implants after a short healing period. A pilot study in minipigs.

The aim of this animal study was to investigate and compare the osseointegration of zirconia and titanium dental implants. 14 one-piece zirconia implants and 7 titanium implants were inserted into the mandibles of 7 minipigs. The zirconia implants were alternately placed submerged and non-submerged. To enable submerged healing, the supraosseous part was removed, using a diamond saw. The titanium implants were all placed submerged. After a healing period of 4 weeks, a histological analysis of the soft and hard tissue and a histomorphometric analysis of the bone-implant contact (BIC) and relative peri-implant bone-volume density (rBVD; relation to bone-volume density of the host bone) was performed. Two zirconia implants were found to be loose. All other implants were available for evaluation. For submerged zirconia and titanium implants, the implant surface showed an intimate connection to the neighbouring bone, with both types achieving a BIC of 53%. For the non-submerged zirconia implants, some crestal epithelial downgrowth could be detected, with a resultant BIC of 48%. Highest rBVD values were found for submerged zirconia (80%), followed by titanium (74%) and non-submerged zirconia (63%). The results suggest that unloaded zirconia and titanium implants osseointegrate comparably, within the healing period studied.

Evaluation of osseointegration of dental implants coated with collagen, chondroitin sulphate and BMP-4: an animal study.

Various studies have shown type I collagen (coll) to increase bone-implant contact (BIC) compared to uncoated implants. The aim of this animal study was to test whether the integration of chondroitin sulphate (CS) and the growth factor rhBMP-4 into a collagenous coating could further increase the measured BIC compared to collagen coated implants alone. The experimental implants had two recesses along the length axis. 120 implants with the surface modifications: coll, coll/CS, coll/CS/rhBMP-4 were inserted into the mandible of 20 minipigs. Six months after implantation, BIC was measured histomorphometrically on the surface and within the recesses. Due to the specific animal model and strict criteria in placement, 39.2 % of the implants were considered as failure and not included in the analysis. Of the successfully gained 73 implants, the highest percentage of BIC was obtained for coll/CS (40%), followed by coll (30%) and coll/CS/rhBMP-4 (27%), P=0.013. BIC within the recesses was highest for coll/CS (51%), followed by coll (43%) and coll/CS/rhBMP-4 (34%), P=0.025. The result suggests that the inclusion of CS slightly increases the BIC compared to collagen coated implants. The further inclusion of a low amount rhBMP-4 had a detrimental effect on bone formation compared to coll/CS, P<0.05.

An experimental study of the biomechanical stability of ultrasound-activated pinned (SonicWeld Rx+Resorb-X) and screwed fixed (Resorb-X) resorbable materials for osteosynthesis in the treatment of simulated craniosynostosis in sheep.

We compared a conventional resorbable screw osteosynthesis with a resorbable, ultrasound-activated pin osteosynthesis, and studied mechanical load capacity and operative handling. This new form of osteosynthesis aims to reduce operation times, and to avoid torque loads and screw fractures to achieve stability. A sheep craniotomy model simulated an operation for dysmorphia on an infant skull. Two rectangular craniotomies of equal size were created in 13 lamb skulls, and each refixed by different means: the first by mesh and 20 screws, and the second by mesh with 20 pins inserted with ultrasound activation. All osteosynthesis material consisted of resorbable amorphous poly-(d,l)-lactide (PDLLA) (Resorb-X, KLS Martin, Tuttlingen, Germany). The insertion time was recorded. The animals were killed at different times, and areas of the healing skull including the plates and pins or screws were removed and divided into sections, which were then tested. In total 74 pin-fixed and 77 screw-fixed samples were obtained. Bending and tensile tests were used to simulate different forms of loading. The time required for the insertion of pins was significantly shorter than for screws. The mechanical tests showed differences in the stability of the bond between the osteosynthesis plate and bone that depended on the osteosynthesis system and the length of time it was in the animal. The pin osteosynthesis gave a stable mechanical load capacity, which was significantly different from that of screw osteosynthesis. Advantages of ultrasound-assisted, resorbable, pin osteosynthesis, include optimum operative handling, reduced insertion time, avoidance of fractures of the fixation elements and higher three-dimensional load capacity.

An experimental in vivo analysis of the resorption to ultrasound activated pins (Sonic weld) and standard biodegradable screws (ResorbX) in sheep.

We compared the healing and reaction in the mandibles of 11 sheep of a conventional bioresorbable screw osteosynthesis with the newly developed ultrasound-activated pin osteosynthesis. The thermal stress caused by insertion of the ultrasound-aided pins leads to no cellular reaction around the pin. There is neither clinical nor histological evidence of any initial inflammation that could have been induced by the insertion. Adequate attachment of fibrous tissue to the pin head and the absence of any inflammation are important preconditions for the introduction of this new method of osteosynthesis into clinical practice. Further advantageous characteristics are easy intraoperative handling and a reduction in operating time, because cutting the thread is not required. There must be sufficient interlinkage of the polymer and the trabecular structures to ensure stability.