Intraoral Imaging Research Articles

Development of an oral cancer detection system through deep learning

ObjectiveWe aimed to develop an AI-based model that uses a portable electronic oral endoscope to capture intraoral images of patients for the detection of oral cancer.Subjects and methodsFrom September 2019 to October 2023, 205 high-quality annotated images of oral cancer were collected using a portable oral electronic endoscope at the Chinese PLA General Hospital for this study. The U-Net and ResNet-34 deep learning models were employed for oral cancer detection. The performance of these models was evaluated using several metrics: Dice coefficient, Intersection over Union (IoU), Loss, Precision, Recall, and F1 Score.ResultsDuring the algorithm model training phase, the Dice values were approximately 0.8, the Loss values were close to 0, and the IoU values were around 0.7. In the validation phase, the highest Dice values ranged between 0.4 and 0.5, while the Loss values increased, and the training loss began to decrease gradually. In the test phase, the model achieved a maximum Precision of 0.96 with a confidence threshold of 0.990. Additionally, with a confidence threshold of 0.010, the highest F1 score reached was 0.58.ConclusionThis study provides an initial demonstration of the potential of deep learning models in identifying oral cancer.

The accuracy of anatomic landmarks on the occlusal plane: a comparative study between conventional and 3D image method

BackgroundTo establish the occlusal plane, the conventional methods for facial analysis to gain accurate alignment of the occlusal plane are inadequate, while 3D technologies are an ideal diagnostic tool. The aims of this research are to compare the difference accuracy of anatomic landmarks on the occlusal plane and ala-tragus line between the conventional clinical method and 3D image method in both non-orthodontic and orthodontic treatment volunteers.MethodsA total of 44 volunteers (22 non-orthodontic and 22 orthodontic treatment volunteers) with normal occlusion were selected. All volunteers received 2 operative methods for occlusal plane determination. In conventional method, the occlusal plane was defined by the fox plane line. The ala-tragus line was defined by the radio-opaque markers. In the 3D image method, the volunteers were recorded intraoral images, 3D facial images and CBCT images. A 3D virtual picture was created using EXOCAD® software. The occlusal plane was generated by the incisal and occlusal surfaces of the teeth. Both methods, the angles and distances between the occlusal plane and ala-tragus line were measured and compared statistically on both sides of each volunteer.ResultsBoth volunteers’ group, the mean angles and distances between the occlusal plane-ala tragus line in the conventional method were reported to be significantly greater than the 3D method (P < 0.05). The percentage difference of angles in conventional method were reported to be significantly higher by 13.61–21.58% (p < 0.05) compared to the 3D method. The percentage difference of distances in the conventional method were reported to be significantly greater than the 3D method by 4.73–7.51% (p < 0.05).ConclusionsWithin the limitation of the study, it can be concluded that both conventional and digital methods for establishing the occlusal plane are not parallel to the occlusal plane. The occlusal plane and ala-tragus line in the conventional method and the 3D method were significantly different in terms of angles and distances in both non-orthodontic and orthodontic treatment volunteers. However, the deviation angle of both methods is approximately 13–20 degrees, which is clinically acceptable for occlusal plane establishment. The accuracy of both methods is still within the using in clinical implementation.

Applications of AI-based deep learning models for detecting dental caries on intraoral images - a systematic review.

This systematic review aimed to assess the effectiveness of ArtificialIntelligence(AI)-based Deep Learning (DL) models in the detection of dental caries on intraoral images. This systematic review adhered to PRISMA 2020 guidelines conducting an electronic search on PubMed, Scopus, and CENTRAL databasesfor retrospective, prospective, and cross-sectional studies published till 1st June 2024. Methodological and performance metrics of clinical studies utilizing DL models were assessed. A modified QUADAS risk of bias tool was used for quality assessment. Out of 273 studies identified, a total of 23 were included with 19 studies having a low risk and 4 studies having a high risk of bias. Overall accuracy ranged from 56% to 99.1%, sensitivity ranged from 23% to 98% and specificity ranged from 65.7% to 100%. Only 3 studies utilized explainable AI (XAI) techniques for caries detection. A total of 4 studies exhibited a level 4 deployment status by developing mobile or web-based applications. AI-based DL models have demonstrated promising prospects in enhancing the detection of dental caries, especially in terms of low-resource settings. However, there is a need for future deployed studies to enhance the AI models to improve their real-world applications.

Oral Cancer Classification with CNN Based State-of-the-art Transfer Learning Methods

The importance of oral and dental health closely affects other vital organs. In this study, CNN-based transfer learning models are built on histopathologic and intraoral images with benign and malignant lesions. Histopathologic and intraoral images from two different sources have benign or malignant classes of lesions in the mouth. EfficientNetB7, ResNet50, VGG16, and VGG19, Xception, ConvNextBase, and MobileNetV2 were used as transfer learning methods. Model training was performed with 80%-20% train test separation and 20% validation separation on the train set. Acc, Prec, Rec, and F1 metrics were used to evaluate the model. In histopathologocial images, Resnet50 was ahead with 81.25% Acc and 85.25% F1. In intraoral images, ConvNextBase with 0.80% Acc, and 0.80% F1 was found to be more accurate.

Recommendations for Research to Develop a Patient-Centered Clinical Follow-Up Protocol for Oral Epithelial Dysplasia.

Oral epithelial dysplasia (OED) is the primary histological marker for assessing the progression of oral potentially malignant disorders (OPMDs) to cancer. Despite challenges in grading and low inter-pathologist reproducibility, OED severity remains the key predictor of malignant transformation. However, globally accepted guidelines for OED monitoring are lacking, despite calls for individualized management based on host and lesion characteristics. The proposed research protocol involves acquiring high-quality intraoral images, assessing oral hygiene and periodontal status, eliminating chronic mechanical irritation and Candida infections, and applying adjunctive diagnostic methods like toluidine blue staining, optical evaluation, and brush cytology. Tailored follow-up regimens based on individual risk assessments are emphasized, with frequent monitoring for high-grade dysplasia or patients at higher risk of progression. Therefore, effective OED management should consider the patient's immune status, dietary habits, and oral microbiota, aiming to develop personalized treatment strategies that optimize patient-centered care.

Developing an AI-based application for caries index detection on intraoral photographs.

This study evaluates the effectiveness of an Artificial Intelligence (AI)-based smartphone application designed for decay detection on intraoral photographs, comparing its performance to that of junior dentists. Conducted at The Aga Khan University Hospital, Karachi, Pakistan, this study utilized a dataset comprising 7,465 intraoral images, including both primary and secondary dentitions. These images were meticulously annotated by two experienced dentists and further verified by senior dentists. A YOLOv5s model was trained on this dataset and integrated into a smartphone application, while a Detection Transformer was also fine-tuned for comparative purposes. Explainable AI techniques were employed to assess the AI's decision-making processes. A sample of 70 photographs was used to directly compare the application's performance with that of junior dentists. Results showed that the YOLOv5s-based smartphone application achieved a precision of 90.7%, sensitivity of 85.6%, and an F1 score of 88.0% in detecting dental decay. In contrast, junior dentists achieved 83.3% precision, 64.1% sensitivity, and an F1 score of 72.4%. The study concludes that the YOLOv5s algorithm effectively detects dental decay on intraoral photographs and performs comparably to junior dentists. This application holds potential for aiding in the evaluation of the caries index within populations, thus contributing to efforts aimed at reducing the disease burden at the community level.

Oral screening of dental calculus, gingivitis and dental caries through segmentation on intraoral photographic images using deep learning

ObjectiveIntraoral photographic images are instrumental in the early screening and clinical diagnosis of oral diseases. In addition, people have been trying to apply artificial intelligence to these images. The purpose of this study is to investigate and evaluate a deep learning system designed to segment intraoral photographic images for the detection of dental caries, dental calculus, and gingivitis, and to assess the degree of dental calculus based on the overall features of the tooth surface and gingival margin.Material and methodsThis cross-sectional study collected 3,365 oral endoscopic images, randomly distributed in training datasets (2,019 images), validation dataset (673 images), and test dataset (673 images). The training set and verification set images are manually labeled. An oral endoscopic image segmentation method based on Mamba (Oral-Mamba) and an intelligent evaluation model of dental calculus degree were proposed, achieving the segmentation of two types of oral diseases, namely gingivitis and dental caries, as well as the segmentation of dental calculus regions, and the intelligent evaluation of the degree of dental calculus.ResultsOral-Mamba demonstrated high accuracy in segmentation, with accuracy rates for gingivitis, dental caries, and dental calculus at 0.83, 0.83, and 0.81, respectively. In particular, these rates surpassed those of the U-Net model in IoU, accuracy, and recall metrics. Furthermore, Oral-Mamba runs 25% faster than U-Net.The accuracy of degree classification in the intelligent evaluation model of dental calculus degree is 85%.ConclusionThe proposed deep learning system is expected to be used for the detection of two types of oral diseases and dental calculus, and the degree judgment of photographic images from an intraoral camera. This system offers a practical method to assist in the oral screening of dental caries, dental calculus, and gingivitis, providing benefits such as intuitive use, time efficiency, cost-effectiveness, and ease of deployment.

For excellent quality intraoral images

For excellent quality intraoral images

Use of a Novel Artificial Intelligence Approach for a Faster and More Precise Computerized Facial Evaluation in Aesthetic Dentistry.

AI is based on automated learning algorithms that use large bodies of information (big data). In the field of dentistry, AI allows the analysis of radiographs, intraoral images and other clinical recordings with unprecedented precision and speed. Facial analysis is known for helping dentists and patients achieve a satisfactory result when a restorative treatment must be realized. The objective of this study is to conduct a neural network-based computerized facial analysis using Python programming language in order to valuate its efficacy in facial point detection. The neural network was trained to identify the main facial and dental points: smile line, lips, size and for of the teeth, etc. A facial analysis was carried out using AI. A descriptive analysis was made with calculation of the mean and standard deviation (SD) of the precision and accuracy in each group. Analysis of variance (ANOVA) was used for the comparison of means between groups. At the intersecting point between dentistry and technology, advances in artificial intelligence (AI) are producing a change in the way modern dentistry is performed. The present study evidenced lesser variability in the execution times of the neural network compared with the DSD system. This indicates that the neural network affords more consistent and predictable results, representing a significant advantage in terms of time and efficacy. The neural network is significantly more efficient and consistent in performing facial analyses than the conventional DSD system. The neural network reduces the time needed to complete the analysis and shows lesser variability in its execution times.

Conventional Intraoral Imaging Exposure Rates among Dental Students

Conventional Intraoral Imaging Exposure Rates among Dental Students

Deep Learning Models for Dental Conditions Classification Using Intraoral Images

This paper presents the digitalization of dentistry medical records to support the dentist in the patient examination process. A dentist uses manual input to fill out the evaluation form by drawing and labeling each patient’s tooth condition based on their observations. Consequently, it takes too long to finish only one examination. For time efficiency, using AI-based digitalization technology can be a promising solution. To address the problem, we made and compared several classification models to recognize human dental conditions to help doctors analyze patient teeth. We apply the YOLOv5, MobileNet V2, and IONet (proposed CNN model) as deep learning models to recognize the five common human dental conditions: normal, filling, caries, gangrene radix, and impaction. We tested the ability of YOLO classification as an object detection model and compared it with classification models. We used a dataset of 3.708 intraoral dental images generated by various augmentation methods from 1.767 original images. We collected and annotated the dataset with the help of dentists. Furthermore, the dataset is divided into three parts: 90% of the total dataset is used as training and validation data, then divided again into 80% training data and 20% validation data. 10% of the total dataset will be used as testing data to compare classification performance. Based on our experiments, YOLOv5, as an object detection model, can classify dental conditions in humans better than the classification model. YOLOv5 produces an 82% accuracy testing value and performs better than the classification model. MobileNet V2 and IONet only get 80% and 70% testing accuracy. Although statistically, there is not much of a difference between the test accuracy values for YOLOv5 and MobileNet v2, the speed in classifying dental objects using YOLOv5 is more efficient, considering that YOLOv5 is an object detection model. There are still challenges with the deep learning technique used in this research, but these can be addressed in further development. A more complex model and the enlargement of more data, ensuring it is varied and balanced, can be used to address the limitations.

DeepPlaq: Dental plaque indexing based on deep neural networks.

The selection of treatment for dental plaque is closely related to the condition of the plaque on different teeth. This study validated the ability of CNN models in assessing the dental plaque indices. In 70 (20 male and 50 female) healthy adults (18 to 55 years old), frontal and lateral view intraoral images (210) of plaque disclosing agent stained permanent and deciduous dentitions were obtained. A three-stage method was employed, where the You Look Only Once version 8 (YOLOv8) model was first used to detect the target teeth, followed by the prompt-based Segment Anything Model (SAM) segmentation algorithm to segment teeth. A new single-tooth dataset consisting of 1400 photographs was obtained after applying a two-stage method. Finally, a multi-class classification model DeepPlaq was trained and evaluated on the accuracy of dental plaque indexing based on the Quigley-Hein Index (QHI) scoring system. Classification performance was measured using accuracy, recall, precision, and F1-score. The teeth detector exhibited an accuracy (mean average precision, mAP) of approximately 0.941 ± 0.005 in identifying teeth with plaque disclosing agents. The maximum accuracy attained in the plaque indexing through DeepPlaq was 0.84 (probability that DeepPlaq scored identical to experts), and the smallest average scoring error was less than 0.25 on a 0 to 5 scale for scoring. A three-stage approach demonstrated excellent performance in detecting and segmenting target teeth, and DeepPlaq model also showed strong performance in assessing dental plaque indices. Application of artificial intelligence to the evaluation of dental plaque distribution could enhance diagnostic accuracy and treatment efficiency and accuracy.

Advances in Artificial Intelligence-aided Intraoral Imaging Analysis in Periodontics

Artificial intelligence has increasingly influenced the field of periodontology by enhancing diagnostic accuracy and treatment planning through advanced data-driven techniques. It was aimed to examine the integration of artificial intelligence, particularly deep learning and machine learning, in analyzing intraoral photographs for periodontal conditions in this review. Periodontal assessments rely on clinical and radiographic evaluations, but artificial intelligence introduces a transformative approach by analyzing large datasets to improve clinical decision-making. The review investigates the effectiveness of artificial intelligence-enhanced intraoral photograph analysis, focusing on methodologies for dataset creation, model development, training, and performance evaluation. A thorough search of databases such as PubMed, Scopus, Google Scholar, and IEEE Xplore identified 338 articles, with 16 meeting the inclusion criteria. These studies primarily utilized convolutional neural networks and architectures like DeepLabv3+ and U-Net, demonstrating high accuracy in detecting conditions such as gingivitis, dental plaque, and other periodontal issues. The dataset sizes ranged from 110 to 7220 images, affecting the models' generalizability. Most studies employed supervised learning, with models trained on labeled datasets to achieve precise diagnostic outcomes. The review highlights that while artificial intelligence and machine learning techniques, including convolutional neural networks and U-Net, offer significant improvements in periodontal diagnostics, the choice of model and the quality of the dataset are crucial for performance. Hybrid approaches that combine automated and expert-driven methods might provide a balance between efficiency and accuracy. The successful integration of artificial intelligence into clinical practice requires continuous validation and adaptation to ensure that these technologies remain accurate and relevant. Future research should focus on enhancing model robustness, expanding dataset diversity, and refining clinical applications to fully exploit the potential of artificial intelligence in periodontology.

Artificial Intelligence for Predicting the Aesthetic Component of the Index of Orthodontic Treatment Need.

The aesthetic component (AC) of the Index of Orthodontic Treatment Need (IOTN) is internationally recognized as a reliable and valid method for assessing aesthetic treatment need. The objective of this study is to use artificial intelligence (AI) to automate the AC assessment. A total of 1009 pre-treatment frontal intraoral photos with overjet values were collected. Each photo was graded by an experienced calibration clinician. The AI was trained using the intraoral images, overjet, and two other approaches. For Scheme 1, the training data were AC 1-10. For Scheme 2, the training data were either the two groups AC 1-5 and AC 6-10 or the three groups AC 1-4, AC 5-7, and AC 8-10. Sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were measured for all approaches. The performance was tested without overjet values as input. The intra-rater reliability for the grader, using kappa, was 0.84 (95% CI 0.76-0.93). Scheme 1 had 77% sensitivity, 88% specificity, 82% accuracy, 89% PPV, and 75% NPV in predicting the binary groups. All other schemes offered poor tradeoffs. Findings after omitting overjet and dataset supplementation results were mixed, depending upon perspective. We have developed deep learning-based algorithms that can predict treatment need based on IOTN-AC reference standards; this provides an adjunct to clinical assessment of dental aesthetics.

Intraoral Ultrasound Imaging Using a Rotational Transducer with Periodontal Feature Identification by Machine Learning.

Innovative intraoral ultrasound devices with smart artificial intelligence-based identification for dento-anatomy could provide crucial information for oral health diagnosis and treatment and shed light on real-time detection of developmental dentistry. However, the grand challenge is that the current ultrasound technologies are meant for external use due to their bulkiness and low frequency. We report a compact versatile ultrasound intraoral device that consists of a rotational probe head robustly pivoted around a hand-held and portable handle for real-time imaging of intraoral anatomy using high-frequency ultrasonography (up to 25 MHz). The intraoral ultrasound device that could be adjusted for various orientations of the imaging planes by rotating the head provides real-time, high-resolution ultrasonograms of intraoral structures, including dento-periodontium of most tooth types and maxillary palate. Machine learning-based algorithms are integrated to automate the identification of important structures, including alveolar bone and cementum-enamel junction. The intraoral ultrasound device smartened with artificial intelligence could innovate oral health diagnosis and treatment plans toward precision health and patient care.

Automatic 3D Teeth Reconstruction From Five Intra-Oral Photos Using Parametric Teeth Model.

Orthodontic treatment is a lengthy process that requires regular in-person dental monitoring, making remote dental monitoring a viable alternative when face-to-face consultation is not possible. In this study, we propose an improved 3D teeth reconstruction framework that automatically restores the shape, arrangement, and dental occlusion of upper and lower teeth from five intra-oral photographs to aid orthodontists in visualizing the condition of patients in virtual consultations. The framework comprises a parametric model that leverages statistical shape modeling to describe the shape and arrangement of teeth, a modified U-net that extracts teeth contours from intra-oral images, and an iterative process that alternates between finding point correspondences and optimizing a compound loss function to fit the parametric teeth model to predicted teeth contours. We perform a five-fold cross-validation on a dataset of 95 orthodontic cases and report an average Chamfer distance of 1.0121 mm2 and an average Dice similarity coefficient of 0.7672 on all the test samples in the cross-validation, demonstrating a significant improvement compared with the previous work. Our teeth reconstruction framework provides a feasible solution for visualizing 3D teeth models in remote orthodontic consultations.

Evaluation of transfer ensemble learning-based convolutional neural network models for the identification of chronic gingivitis from oral photographs

BackgroundTo evaluate the performances of several advanced deep convolutional neural network models (AlexNet, VGG, GoogLeNet, ResNet) based on ensemble learning for recognizing chronic gingivitis from screening oral images.MethodsA total of 683 intraoral clinical images acquired from 134 volunteers were used to construct the database and evaluate the models. Four deep ConvNet models were developed using ensemble learning and outperformed a single model. The performances of the different models were evaluated by comparing the accuracy and sensitivity for recognizing the existence of gingivitis from intraoral images.ResultsThe ResNet model achieved an area under the curve (AUC) value of 97%, while the AUC values for the GoogLeNet, AlexNet, and VGG models were 94%, 92%, and 89%, respectively. Although the ResNet and GoogLeNet models performed best in classifying gingivitis from images, the sensitivity outcomes were not significantly different among the ResNet, GoogLeNet, and Alexnet models (p>0.05). However, the sensitivity of the VGGNet model differed significantly from those of the other models (p < 0.001).ConclusionThe ResNet and GoogLeNet models show promise for identifying chronic gingivitis from images. These models can help doctors diagnose periodontal diseases efficiently or based on self-examination of the oral cavity by patients.

Long-term prognosis for teeth with external cervical resorption based on periapical images and cone beam CT: A clinical study.

To evaluate whether supplemental information from CBCT changed long-term prognosis for teeth with external cervical resorption (ECR) compared to periapical (PA) images. Furthermore, to assess predictive findings in PA images and evaluate which findings in CBCT affected the long-term prognosis of teeth with ECR. One hundred and ninty-four patients, mean age 41.2, range 13-81, having 244 teeth with ECR were included. An initial long-term prognosis determined either good or poor was established based on intraoral images. Afterwards, the patients underwent CBCT, and final long-term prognosis was decided. From the PA images and CBCT, ECR using Heithersay's classification system, pulp involvement and extension of ECR was assessed. In CBCT, the number of surface lesion(s) was additionally assessed. Descriptive statistics evaluated changes in long-term prognosis after CBCT. Logistic regression analyses tested if findings in PA images and CBCT affected the long-term prognosis. Based on CBCT, out of 244 teeth the long-term prognosis was assessed to be poor for 173 (70.9%) teeth and good for 71 (29.1%) teeth. The long-term prognosis changed in 76 (31.1%) teeth after CBCT; 5 (2.0%) changed from poor to good, and 71 (29%) changed from good to poor long-term prognosis. In 81 (33.2%) teeth the H-class increased, and in 10 (4.1%) teeth the H-class decreased after assessing CBCT. In 70 (28.7%) teeth, there was a change from no pulp involvement to involvement of the pulp after CBCT; eight (3.3%) teeth changed in the opposite direction. H-class 2 and 4 in PA images significantly increased the probability for a change in long-term prognosis compared to an H-class 3 (p < .05). H-class 4, pulp involvement, ECR in the oral 1/3 of the root, and more than two surface lesions seen in CBCT significantly influenced a poor long-term prognosis (p < .05). Supplemental information from CBCT changed long-term prognosis in almost one third of teeth with ECR. In most cases, the long-term prognosis changed from good to poor. H-class 3 in PA images had a significant influence on change in long-term prognosis. Several findings in CBCT influenced a poor long-term prognosis.

A deep learning approach to detection of oral cancer lesions from intra oral patient images: A preliminary retrospective study

IntroductionOral squamous cell carcinomas (OSCC) seen in the oral cavity are a category of diseases for which dentists may diagnose and even cure. This study evaluated the performance of diagnostic computer software developed to detect oral cancer lesions in intra-oral retrospective patient images. Materials and methodsOral cancer lesions were labeled with CranioCatch labeling program (CranioCatch, Eskişehir, Turkey) and polygonal type labeling method on a total of 65 anonymous retrospective intraoral patient images of oral mucosa that were diagnosed with oral cancer histopathologically by incisional biopsy from individuals in our clinic. All images have been rechecked and verified by experienced experts. This data set was divided into training (n = 53), validation (n = 6) and test (n = 6) sets. Artificial intelligence model was developed using YOLOv5 architecture, which is a deep learning approach. Model success was evaluated with confusion matrix. ResultsWhen the success rate in estimating the images reserved for the test not used in education was evaluated, the F1, sensitivity and precision results of the artificial intelligence model obtained using the YOLOv5 architecture were found to be 0.667, 0.667 and 0.667, respectively. ConclusionsOur study reveals that OCSCC lesions carry discriminative visual appearances, which can be identified by deep learning algorithm. Artificial intelligence shows promise in the prediagnosis of oral cancer lesions. The success rates will increase in the training models of the data set that will be formed with more images.

Caries lesions diagnosis with deep convolutional neural network in intraoral QLF images by handheld device

ObjectivesThis study investigated the effectiveness of a deep convolutional neural network (CNN) in diagnosing and staging caries lesions in quantitative light-induced fluorescence (QLF) images taken by a self-manufactured handheld device.MethodsA small toothbrush-like device consisting of a 400 nm UV light-emitting lamp with a 470 nm filter was manufactured for intraoral imaging. A total of 133 cases with 9,478 QLF images of teeth were included for caries lesion evaluation using a CNN model. The database was divided into development, validation, and testing cohorts at a 7:2:1 ratio. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and area under the receiver operating characteristic curve (AUC) were calculated for model performance.ResultsThe overall caries prevalence was 19.59%. The CNN model achieved an AUC of 0.88, an accuracy of 0.88, a specificity of 0.94, and a sensitivity of 0.64 in the validation cohort. They achieved an overall accuracy of 0.92, a sensitivity of 0.95 and a specificity of 0.55 in the testing cohort. The model can distinguish different stages of caries well, with the best performance in detecting deep caries followed by intermediate and superficial lesions.ConclusionsCaries lesions have typical characteristics in QLF images and can be detected by CNNs. A QLF-based device with CNNs can assist in caries screening in the clinic or at home.Trial registrationThe clinical trial was registered in the Chinese Clinical Trial Registry (No. ChiCTR2300073487, Date: 12/07/2023).