Teeth segmentation and carious lesions segmentation in panoramic X-ray images using CariSeg, a networks’ ensemble

Background: Dental cavities are common oral diseases that can lead to pain, discomfort, and eventually, tooth loss. Early detection and treatment of cavities can prevent these negative consequences. We propose CariSeg, an intelligent system composed of four neural networks that result in the detection of cavities in dental X-rays with 99.42% accuracy.

Method: The first model of CariSeg, trained using the U-Net architecture, segments the area of interest, the teeth, and crops the radiograph around it. The next component segments the carious lesions and it is an ensemble composed of three architectures: U-Net, Feature Pyramid Network, and DeeplabV3. For tooth identification two merged datasets were used: The Tufts Dental Database consisting of 1000 panoramic radiography images and another dataset of 116 anonymized panoramic X-rays, taken at Noor Medical Imaging Center, Qom. For carious lesion segmentation, a dataset consisting of 150 panoramic X-ray images was acquired from the Department of Oral and Maxillofacial Surgery and Radiology, Iuliu Hatieganu University of Medicine and Pharmacy, Cluj-Napoca.

Results: The experiments demonstrate that our approach results in 99.42% accuracy and a mean 68.2% Dice coefficient. Conclusions: AI helps in detecting carious lesions by analyzing dental X-rays and identifying cavities that might be missed by human observers, leading to earlier detection and treatment of cavities and resulting in better oral health outcomes.

Published in

Mărginean AC, Mureşanu S, Hedeşiu M, Dioşan L. Teeth segmentation and carious lesions segmentation in panoramic X-ray images using CariSeg, a networks’ ensemble. Heliyon. 2024 May 10;10(10):e30836. doi: 10.1016/j.heliyon.2024.e30836. PMID: 38803980; PMCID: PMC11128823.

---

Short-term aesthetic rehabilitation with 3d printed snap-on smile devices – literature review and a case report

Aim of the study: The main objective of this paper is to perform an up-to-date literature review of the application and implications of using Snap-On Smile devices for short-term aesthetic rehabilitation, as well as to showcase the 3D printing workflow in manufacturing these devices through a case study.

Materials and methods: The present systematic study was conducted following the PRISMA-P structure of a systematic review, by using three electronic databases (PubMed, Google Scholar, and EMBASE) to perform a literature search from 2000 to 2024 using the following MESH terms: snap on smile, aesthetic, removable, device. After discarding duplicates, out of 81 results, a total of 6 articles were eligible and included in the review.

Results: Constructed from durable materials like crystallized acetyl resin or PMMA, the Snap-On Smile offers up to five years of use. Manufactured through advanced techniques such as injection moulding, CAD/CAM milling, or 3D printing, it provides an aesthetic solution without necessitating tooth preparation. The case study of a 59-year-old female illustrates its efficacy in providing immediate aesthetic enhancement for a significant social event.

Conclusions: The Snap-On Smile has emerged as a significant advancement in aesthetic dentistry, balancing the desire for improved dental aesthetics with the need to preserve natural tooth structures. Its ease of application and non-invasive nature make it a suitable option for many patients, though its limitations necessitate careful consideration in complex cases.

Published in

Burde AV, Frățilă C, Varvară EB, Varvară AV, Short-term aesthetic rehabilitation with 3d printed snap-on smile devices – literature review and a case report, Romanian Journal of Oral Rehabilitation, Vol. 16, No.2 April-June 2024, DOI: 10.6261/RJOR.2024.2.16.58

---

Automating Dental Condition Detection on Panoramic Radiographs: Challenges, Pitfalls, and OpportunitiesAutomating Dental Condition Detection on Panoramic Radiographs: Challenges, Pitfalls, and Opportunities

Background/Objectives: The integration of AI into dentistry holds promise for improving diagnostic workflows, particularly in the detection of dental pathologies and pre-radiotherapy screening for head and neck cancer patients. This study aimed to develop and validate an AI model for detecting various dental conditions, with a focus on identifying teeth at risk prior to radiotherapy. 

Methods: A YOLOv8 model was trained on a dataset of 1628 annotated panoramic radiographs and externally validated on 180 radiographs from multiple centers. The model was designed to detect a variety of dental conditions, including periapical lesions, impacted teeth, root fragments, prosthetic restorations, and orthodontic devices. 

Results: The model showed strong performance in detecting implants, endodontic treatments, and surgical devices, with precision and recall values exceeding 0.8 for several conditions. However, performance declined during external validation, highlighting the need for improvements in generalizability. 

Conclusions: YOLOv8 demonstrated robust detection capabilities for several dental conditions, especially in training data. However, further refinement is needed to enhance generalizability in external datasets and improve performance for conditions like periapical lesions and bone loss.

Published in

Mureșanu S, Hedeșiu M, Iacob L, Eftimie R, Olariu E, Dinu C, Jacobs R, On Behalf Of Team Project Group. Automating Dental Condition Detection on Panoramic Radiographs: Challenges, Pitfalls, and Opportunities. Diagnostics (Basel). 2024 Oct 21;14(20):2336. doi: 10.3390/diagnostics14202336. PMID: 39451659; PMCID: PMC11507083.

---

A Vision-Guided Robotic System for Safe Dental Implant Surgery

Background: Recent advancements in dental implantology have significantly improved outcomes, with success rates of 90-95% over a 10-year period. Key improvements include enhanced preplanning processes, such as precise implant positioning, model selection, and optimal insertion depth. However, challenges remain, particularly in achieving correct spatial positioning and alignment of implants for optimal occlusion. These challenges are pronounced in patients with reduced bone substance or complex anthropometric features, where even minor misalignments can result in complications or defects. 

Methods: This paper introduces a vision-guided robotic system designed to improve spatial positioning accuracy during dental implant surgery. The system incorporates advanced force-feedback control to regulate the pressure applied to bone, minimizing the risk of bone damage. A preoperative CBCT scan, combined with real-time images from a robot-mounted camera, guides implant positioning. A personalized marker holder guide, developed from the initial CBCT scan, is used for patient-robot calibration. The robot-mounted camera provides continuous visual feedback of the oral cavity during surgery, enabling precise registration of the patient with the robotic system. 

Results: Initial experiments were conducted on a 3D-printed mandible using a personalized marker holder. Following successful patient-robot registration, the robotic system autonomously performed implant drilling. To evaluate the accuracy of the robotic-assisted procedure, further tests were conducted on 40 identical molds, followed by measurements of implant positioning. The results demonstrated improved positioning accuracy compared to the manual procedure. 

Conclusions: The vision-guided robotic system significantly enhances the spatial accuracy of dental implants compared to traditional manual methods. By integrating advanced force-feedback control and real-time visual guidance, the system addresses key challenges in implant positioning, particularly for patients with complex anatomical structures. These findings suggest that robotic-assisted implant surgery could offer a safer and more precise alternative to manual procedures, reducing the risk of implant misalignment and associated complications.

Published in

Pisla D, Bulbucan V, Hedesiu M, Vaida C, Zima I, Mocan R, Tucan P, Dinu C, Pisla D, Team Project Group. A Vision-Guided Robotic System for Safe Dental Implant Surgery. J Clin Med. 2024 Oct 23;13(21):6326. doi: 10.3390/jcm13216326. PMID: 39518466; PMCID: PMC11546886.

---

Real-Time Motion Compensation for Dynamic Dental Implant Surgery

Background: Accurate and stable instrument positioning is critical in dental implant procedures, particularly in anatomically constrained regions. Conventional navigation systems assume a static patient head, limiting adaptability in dynamic surgical conditions. This study proposes and validates a real-time motion compensation framework that integrates optical motion tracking with a collaborative robot to maintain tool alignment despite patient head movement. 

Methods: A six-camera OptiTrack Prime 13 system tracked rigid markers affixed to a 3D-printed human head model. Real-time head pose data were streamed to a Kuka LBR iiwa robot, which guided the implant handpiece to maintain alignment with a predefined target. Motion compensation was achieved through inverse trajectory computation and second-order Butterworth filtering to approximate realistic robotic response. Controlled experiments were performed using the MAiRA Pro M robot to impose precise motion patterns, including pure rotations (±30° at 10–40°/s), pure translations (±50 mm at 5–30 mm/s), and combined sinusoidal motions. Each motion profile was repeated ten times to evaluate intra-trial repeatability and dynamic response. 

Results: The system achieved consistent pose tracking errors below 0.2 mm, tool center point (TCP) deviations under 1.5 mm across all motion domains, and an average latency of ~25 ms. Overshoot remained minimal, with effective damping during motion reversal phases. The robot demonstrated stable and repeatable compensation behavior across all experimental conditions. 

Conclusions: The proposed framework provides reliable real-time motion compensation for dental implant procedures, maintaining high positional accuracy and stability in the presence of head movement. These results support its potential for enhancing surgical safety and precision in dynamic clinical environments.

Published in

Pisla, D.; Bulbucan, V.; Hedeșiu, M.; Vaida, C.; Cailean, A.; Mocan, R.; Tucan, P.; Dinu, C.; Pisla, D.; TEAM Project Group. Real-Time Motion Compensation for Dynamic Dental Implant Surgery. J. Clin. Med. 2025, 14, 6429. https://doi.org/10.3390/jcm14186429

---

Tooth-level detection and mapping of dental pathologies on panoramic radiographs using YOLOv11 and RT-DETR

This article describes a reproducible method for automated tooth-level detection and mapping of dental pathologies on panoramic radiographs. The workflow integrates two deep learning stages: (1) segmentation of individual teeth and (2) detection and classification of dental lesions. Outputs are combined to assign pathologies to specific teeth, supporting comprehensive radiological assessment. The method was implemented and tested using two datasets: a collection of 1,628 panoramic radiographs annotated with 12 pathology categories, and the public Tufts Dental Database. Several model architectures were evaluated, including YOLOv11 and RT-DETR configurations. The influence of image formatting and preprocessing strategies on performance was also assessed. Results showed that RT-DETR-x achieved the highest mean average precision (mAP@50) for pathology detection, while YOLOv11x produced the most accurate tooth segmentation. The integrated system demonstrated high precision in linking lesions to the corresponding teeth. This method offers a practical framework for developing AI-assisted diagnostic tools in dentistry and can be adapted to other imaging datasets. • Development of a fully automated dental diagnostic tool that segments individual teeth and maps detected pathologies to their corresponding tooth • YOLOv11 and RT-DETR variants are extensively tested for segmentation and detection tasks • Preprocessing and image formatting strategies are systematically compared.

Published in

Eftimie S, Ileni T, Iacob L, Hedeșiu M, Dioșan L. Tooth-level detection and mapping of dental pathologies on panoramic radiographs using YOLOv11 and RT-DETR. MethodsX. 2025 Oct 28;15:103696. doi: 10.1016/j.mex.2025.103696. PMID: 41277983; PMCID: PMC12637385.

---

Pioneering fully automated bony orbit segmentation: an in silico nnU-Net multicentre approach

In maxillofacial surgery, orbital reconstruction requires precision to address both functional and aesthetic considerations arising from both acute and elective conditions. This study presents a novel, fully automated segmentation software designed specifically for the orbital floor. This software enhances surgical planning through superior accuracy, efficiency, and usability. The authors’ transdisciplinary team compiled a dataset of 1004 expert-segmented orbits from computed tomography images across multiple countries, ensuring broad anatomical representation. Developed with the nnU-Net framework, the software achieved segmentation accuracy with a mean Dice similarity coefficient of 0.935 and a mean surface distance of 0.292 mm for the orbit, and a Dice similarity coefficient of 0.917 and mean surface distance of 0.287 mm for the orbital floor-all within approximately 1 s. This performance surpasses traditional manual segmentation, which averages 25 min per orbit. The system delivered consistent results across a range of imaging sources, affirming its reliability for a wide range of clinical applications. By introducing this fully automated, high-precision tool, this study pioneers advancements in AI-driven orbital reconstructions, setting new standards for patient-specific surgical planning. Further development and integration holds the key to transforming the field, ensuring ethical compliance and enhancing informed consent.

Published in

Rashad A, Beyer M, Eftimie S, Hinrichs-Puladi B, Vladu O, Xie K, Truhn D, Thieringer FM, Gander T, Hölzle F, Egger J, Ilesan RR. Pioneering fully automated bony orbit segmentation: an in silico nnU-Net multicentre approach. Int J Oral Maxillofac Surg. 2026 May;55(5):542-548. doi: 10.1016/j.ijom.2025.11.002. Epub 2025 Nov 22. PMID: 41274846.

---

ConvU-NExT: An Asymmetrical Encoder–Decoder for Denoising Low Dose CT

Low-dose computed tomography (LDCT) is a medical imaging modality designed to minimize ionizing radiation exposure while maintaining the ability to produce detailed cross-sectional images. It is particularly valuable in scenarios requiring repeated imaging, such as cancer screening, follow-up examinations or pediatric diagnostics, where reducing radiation dose is critical to patient safety. For example, to reduce noise by half, four times the radiation dose is required in the slice. The goal is to achieve post-processed LDCT images with comparable quality to those obtained from standard-dose CT imaging. We start with a brief overview of the CT procedures and their limitations. Then we introduce a novel denoising method based on an asymmetric integration of the ConvNeXt backbone with the U-Net architecture. This novel approach obtained 2–3 times less noise than the original LDCT, having a 10%–20% increase in performance compared to U-Net implementation, checked against three metrics MSE, SSIMLoss and combinations of both. The results suggest that: (i) augmenting the images with specific noise, obtained from water phantom CT scan test, while training yields superior results compared to generic noise augmentations; (ii) a larger kernel size better extracts features and (iii) a smaller kernel size was mandatory for feature reconstruction.

Published in

Iacob LM, Oniţ RD, Groza A, Varan A, Mureșanu S, Hedeșiu M, et al. ConvU-NExT: An asymmetrical encoder–decoder for denoising low dose CT. IET Image Process. 2026;20(1):e70317. doi:10.1049/ipr2.70317

---

The Effect of Nozzle Diameter on the Dimensional Accuracy of FDM-Printed Dental Models

Additive manufacturing has become an integral component of the digital workflow in dentistry. Among available technologies, fused deposition modeling (FDM) offers a cost-effective alternative for producing dental models, although concerns remain regarding dimensional accuracy. The aim of this study was to evaluate the influence of nozzle diameter and dental arch morphology on the dimensional accuracy of FDM-printed dental models. Reference STL datasets representing ideal maxillary and mandibular arches were printed using a Bambu Lab X1 Carbon printer with two nozzle diameters (0.2 mm and 0.4 mm) in a 2 × 2 factorial design (n = 40). Dimensional accuracy was assessed using Root Mean Square (RMS) deviation obtained through 3D surface comparison. Models printed with the 0.2 mm nozzle demonstrated significantly lower RMS deviations compared with those printed using the 0.4 mm nozzle. Additionally, mandibular models exhibited higher dimensional stability than maxillary models. These findings indicate that smaller nozzle diameters can improve the accuracy of FDM-printed dental models.

Published in

Manole MC, Burde AV, Grecu A, Hedeșiu M, Sinescu C. The effect of nozzle diameter on the dimensional accuracy of FDM-printed dental models. Med Evol. 2026;32(1). doi:10.70921/medev.v32i1.2030

---

AI-Assisted Identification of the Medial Lingual Foramen on CBCT: A Deep Learning Approach for Preoperative Implant Assessment

Background and Objectives: Although the anterior mandible is generally considered a safe region for implant placement, injury to the medial lingual foramen (MLF) may result in significant vascular complications. Accurate identification of this structure is challenging due to its small size, low volumetric representation, and anatomical variability. This study aimed to evaluate the anatomical characteristics of the MLF using cone-beam computed tomography (CBCT) and to develop and validate a deep learning-based approach for its automated detection and segmentation. 

Materials and Methods: A total of 106 CBCT scans were retrospectively analyzed to assess the morphology and position of the MLF. Manual pixel-wise annotations of the complete canal trajectory were performed on sagittal slices and used to train convolutional neural network models based on a U-Net-derived framework. Multiple configurations, including multi-class, binary, two-dimensional, and three-dimensional approaches, were evaluated. Given the extremely limited volumetric representation of the MLF, severe class imbalance represented a major challenge during model training and evaluation. Model performance was assessed using the Dice similarity coefficient, precision, recall, and Hausdorff distance. External validation was performed on an independent dataset of 10 CBCT scans. 

Results: The MLF was identified in all patients, with a single canal observed in 63% of cases. The sagittal-plane binary segmentation model achieved the best performance, with a test Dice score of 0.79, precision of 0.88, and recall of 0.73. External validation demonstrated a Dice score of 0.81, precision of 0.89, and recall of 0.71. The 95th percentile Hausdorff distance was 2.6 mm, and the mean center-point localization error was 1.2 mm. The model correctly detected the MLF in 90% of external cases. 

Conclusions: Deep learning-based segmentation of the MLF is feasible and may support automated localization assistance during preoperative CBCT assessment. Performance was influenced by the alignment between the annotation strategy and model input, highlighting an important consideration for small-structure segmentation. Further validation on larger multicenter datasets is required before clinical implementation can be considered.

Published in

Ban, A., Mureşanu, S., Roman, R., Iacob, L., Hedeşiu, M., Dinu, C., Almăşan, O., & on behalf of Team Project Group. (2026). AI-Assisted Identification of the Medial Lingual Foramen on CBCT: A Deep Learning Approach for Preoperative Implant Assessment. Medicina, 62(6), 1059. https://doi.org/10.3390/medicina62061059

---

COMPARISON OF DIMENSIONAL ACCURACY BETWEEN PROFESSIONAL DLP AND CONSUMER LCD 3D PRINTERS FOR DENTAL MODEL FABRICATION

Aim of the study: The aim of this study was to compare the dimensional accuracy of dental models fabricated using a medical-grade DLP printer and a consumer-grade LCD printer. Materials and methods: Standardized maxillary and mandibular STL models were printed using a DLP system (Bego Varseo S) and an LCD system (Phrozen Mini 8K S). A total of 20 models were digitized and compared with the reference files using threedimensional surface deviation analysis. Dimensional trueness was assessed using root mean square (RMS) and squared mean surface distance (SMSD) metrics, followed by non-parametric statistical analysis. Results: Both systems produced comparable dimensional accuracy, with no statistically significant differences in RMS or SMSD values between groups. Maxillary models showed higher deviations than mandibular models for both technologies. A strong correlation was observed between RMS and SMSD values. Conclusions: Within the limitations of this study, consumer-grade LCD printing demonstrated dimensional accuracy comparable to that of a medical-grade DLP system for dental model fabrication.

Published in

Alexandru-Victor Burde, Michel Beyer, Marius Ciprian Manole, Rareș Mario Borșa, Sabina-Gabriela Crețu, Mihaela Hedeșiu, Cosmin Sinescu. Comparison Of Dimensional Accuracy Between Professional Dlp And Consumer Lcd 3d Printers For Dental Model Fabrication. Romanian Journal of Oral Rehabilitation Vol. 18, No.1 January-March 2026. DOI: 10.62610/RJOR.2025.1.18.9