Hybrid CNN-Transformer Model for Accurate Impacted Tooth Detection in Panoramic Radiographs

Diagnostics, Journal Year: 2025, Volume and Issue: 15(3), P. 244 - 244

Published: Jan. 22, 2025

Background/Objectives: The integration of digital imaging technologies in dentistry has revolutionized diagnostic and treatment practices, with panoramic radiographs playing a crucial role detecting impacted teeth. Manual interpretation these images is time consuming error prone, highlighting the need for automated, accurate solutions. This study proposes an artificial intelligence (AI)-based model teeth radiographs, aiming to enhance accuracy reliability. Methods: proposed combines YOLO (You Only Look Once) RT-DETR (Real-Time Detection Transformer) models leverage their strengths real-time object detection learning long-range dependencies, respectively. further optimized Weighted Boxes Fusion (WBF) algorithm, where WBF parameters are tuned using Bayesian optimization. A dataset 407 labeled was used evaluate model’s performance. Results: achieved mean average precision (mAP) 98.3% F1 score 96%, significantly outperforming individual other combinations. results were expressed through key performance metrics, such as mAP scores, which highlight balance between recall. Visual numerical analyses demonstrated superior performance, enhanced sensitivity minimized false positive rates. Conclusions: presents scalable reliable AI-based solution offering substantial improvements efficiency. potential widespread application clinical dentistry, reducing manual workload Future research will focus on expanding refining generalizability.

Language: Английский

---

Augmented Intelligence in Oral and Maxillofacial Radiology: A Systematic Review

Oral Surgery Oral Medicine Oral Pathology and Oral Radiology, Journal Year: 2025, Volume and Issue: unknown

Published: April 1, 2025

Artificial intelligence (AI) is transforming diagnostic imaging in dentistry. This systematic review evaluates existing literature on augmented dentomaxillofacial radiology, focusing its influence human collaboration interpreting dental imaging. A search across seven databases and gray was conducted. Studies evaluating clinician performance with AI-assistance were included, while reviews, surveys, case reports excluded. The QUADAS-2 tool assessed the risk of bias. Sixteen studies AI radiographic interpretation. AI-assisted caries detection consistently improved accuracy, sensitivity, specificity. Detection apical pathoses jaw lesion segmentation reducing time. Cephalometric landmark identification showed increased particularly for students. Soft tissue calcification but sensitivity decreased. Overall, enhanced interobserver agreement reduced variability, general dentists students showing greatest gains. Augmented enhances interpretation by improving tasks, less experienced clinicians, positively influences clinical decision-making. However, remains inconsistent challenging cases involving complex or varied conditions. While it complements rather than replaces further validation AI's generalizability reliability using larger, diverse datasets necessary.

Language: Английский

---

Artificial Intelligence in Detecting and Segmenting Vertical Misfit of Prosthesis in Radiographic Images of Dental Implants: A Cross‐Sectional Analysis

Clinical Oral Implants Research, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 22, 2025

This study evaluated ResNet-50 and U-Net models for detecting segmenting vertical misfit in dental implant crowns using periapical radiographic images. Periapical radiographs of were classified by two experts based on the presence (reference group). The area was manually annotated images exhibiting misfit. resulting datasets utilized to train deep learning models. Then, 70% allocated training, while remaining 30% used validation testing. Five general dentists categorized testing as "misfit" or "fit." Inter-rater reliability with Cohen's kappa index performance metrics calculated. average artificial intelligence (AI) compared paired-samples t test. A total 638 collected. values between AI ranged from 0.93 0.98, indicating perfect agreement. model achieved accuracy precision 92.7% 87.5%, respectively, whereas had a mean 93.3% 89.6%. sensitivity specificity 90.3% 93.8%, 90.1% 95.1% dentists. Dice coefficient yielded 88.9% 89.5% among algorithm produced loss 0.01 an 0.98. No significant difference found (p > 0.05). can detect segment prosthetic radiographs, comparable clinician performance.

Language: Английский

---

AI-Driven Dental Caries Management Strategies: From Clinical Practice to Professional Education and Public Self Care

International Dental Journal, Journal Year: 2025, Volume and Issue: 75(4), P. 100827 - 100827

Published: May 10, 2025

Language: Английский

---

PDCNET: Deep Convolutional Neural Network for Classification of Periodontal Disease using Dental Radiographs

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 150147 - 150168

Published: Jan. 1, 2024

Language: Английский

---

Clinical Validation of Deep Learning for Segmentation of Multiple Dental Features in Periapical Radiographs

Bioengineering, Journal Year: 2024, Volume and Issue: 11(10), P. 1001 - 1001

Published: Oct. 5, 2024

Periapical radiographs are routinely used in dental practice for diagnosis and treatment planning purposes. However, they often suffer from artifacts, distortions, superimpositions, which can lead to potential misinterpretations. Thus, an automated detection system is required overcome these challenges. Artificial intelligence (AI) has been revolutionizing various fields, including medicine dentistry, by facilitating the development of intelligent systems that aid performing complex tasks such as planning. The purpose present study was verify diagnostic performance AI automatic teeth, caries, implants, restorations, fixed prosthesis on periapical radiographs. A dataset comprising 1000 collected 500 adult patients analyzed compared with annotations provided two oral maxillofacial radiologists. strong correlation (R > 0.5) observed between perception observers 1 2 carious teeth (0.7-0.73), implants (0.97-0.98), restored (0.85-0.89), (0.92-0.94), missing (0.82-0.85). comparable radiologists may be useful identification

Language: Английский

---

Enhancing dental caries classification in CBCT images by using image processing and self-supervised learning

Computers in Biology and Medicine, Journal Year: 2024, Volume and Issue: 183, P. 109221 - 109221

Published: Oct. 7, 2024

Language: Английский

---

Integração de processamento de imagens e aprendizado de máquina na identificação de cáries em imagens de tomografia.

Published: Aug. 1, 2024

Os exames de Tomografia Computadorizada Feixe Cônico (TCFC) são essenciais na área Odontologia, tanto para o diagnóstico preciso quanto planejamento tratamentos dentários complexos.Um dos achados possíveis nas imagens TCFC as suspeitas cáries, que constituem um desafio por ser difícil a sua diferenciação artefatos, principalmente causados pela atenuação materiais densos ou metálicos.Outra dificuldade se encontra identificação cáries em estágios iniciais, cujo tamanho dificulta visualização TCFC.A e classificação meio desses possibilita intervenções precoces eficazes lesões cariosas.No entanto, abordagem tradicional dessas depende fortemente avaliações visuais interpretações especialistas, processo exige tempo esforço.As metodologias segmentação técnicas Aprendizado Máquina têm sido cada vez mais adotadas Odontologia aprimorar outras doenças.A partir uma revisão sistemática métodos computacionais foi constatada notável ausência diferenciem os das importante definir tratamento da mesma.Considerando avanços tecnológicos recentes tendências limitações identificadas literatura atual, objetivo deste projeto é desenvolver método dentárias TCFC, integrando processamento com Máquina.Adicionalmente, incorpora uso Profundo conjunto tarefas Self-supervised Learning, empregando escala do International Caries Detection and Assessment System diferentes níveis cáries.Na etapa segmentação, atingido segmentar automaticamente região lesionada, precisão 88,50%, sensibilidade 87,33% índice Jaccard 0,7037.Na classificação, foram utilizadas duas abordagens: Clássico Learning (SSL).Na Clássico, modelos apresentaram acurácia F1-Score macro acima 86%.Utilizando SSL, superior 88%, representando melhoria significativa relação aos anteriores.Esses resultados mostram eficácia proposta ao apresentar análise abordagens e, consequentemente, afetada

---

Development and evaluation of a deep learning segmentation model for assessing non-surgical endodontic treatment outcomes on periapical radiographs: A retrospective study

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(12), P. e0310925 - e0310925

Published: Dec. 31, 2024

This study aimed to evaluate the performance of a deep learning-based segmentation model for predicting outcomes non-surgical endodontic treatment. Preoperative and 3-year postoperative periapical radiographic images each tooth from routine root canal treatments performed by endodontists 2015 2021 were obtained retrospectively Thammasat University hospital. 1200 teeth with follow-up results (440 healed, 400 healing, 360 disease) collected. Mask Region-based Convolutional Neural Network (Mask R-CNN) was used pixel-wise segment other structures in image trained predict class label into healing disease. Three annotated 1080 training, validation, testing. The evaluated on test set also comparison clinicians (general practitioners endodontists) without help independent 120 images. R-CNN prediction high mean average precision (mAP) 0.88 (95% CI 0.83–0.93) area under precision-recall curve 0.91 0.88–0.94), 0.83 0.81–0.85), 0.90–0.92) disease, respectively. metrics general significantly improved outperforming alone mAP increasing 0.75 0.72–0.78) 0.84 0.81–0.87) 0.85–0.91) 0.92 0.89–0.95), In conclusion, had potential treatment expected aid

Language: Английский

---

A Systematic Review of the Impact of Artificial Intelligence (AI) on Dental Diagnosis

Advances in medical technologies and clinical practice book series, Journal Year: 2024, Volume and Issue: unknown, P. 21 - 46

Published: Dec. 31, 2024

This chapter systematically reviews the impact of Artificial Intelligence (AI) on dental diagnosis, focusing its applications in digital dentistry, particularly rural areas where access to quality care is limited. We conducted a review literature examining research articles, medical trials and case reports that explore use intelligence tools, like Google Cloud Vision API Vertex AI diagnosis. The looked into factors such as accuracy effectiveness impact, patients results. findings suggests powered diagnostic systems greatly improve efficiency diagnoses detecting cavities, gum diseases other oral health issues. Utilizing cloud based services offers options, for clinics areas, with limited resources encouraging early detection tailored treatment strategies. Despite challenges associated data privacy need model validation

Language: Английский

---