An automated dental caries detection and scoring system for optical images of tooth occlusal surface

Observational Diagnostics: The Building Block of AI-Powered Visual Aid for Dental Practitioners

Artificial intelligence (AI) has gained significant traction in medical image analysis, including dentistry, aiding clinicians in making timely and accurate diagnoses. Radiographs, such as orthopantomograms (OPGs) and intraoral radiographs, along with clinical photographs, are the primary imaging modalities employed for AI-powered analysis in the dental field. In this review, we discuss the most recent research and product developments concerning the clinical application of AI as a visual aid in dentistry and introduce the concept of Observational Diagnostics (ODs) as a structured method to standardise image analysis. ODs serve as foundational elements for AI-driven diagnostic aids and have the potential to improve the consistency and reliability of diagnostic data used in treatment planning. We provide illustrative examples to demonstrate how ODs not only represent a significant advancement towards more precise diagnostic aids but also provide the basis for the generation of evidence-based treatment recommendations. These OD-based algorithms have been integrated into chairside AI applications to streamline clinical workflows to improve consistency, accuracy, and efficiency.

Detecting dental caries on oral photographs using artificial intelligence: A systematic review

OBJECTIVES

This systematic review aimed at evaluating the performance of artificial intelligence (AI) models in detecting dental caries on oral photographs.

METHODS

Methodological characteristics and performance metrics of clinical studies reporting on deep learning and other machine learning algorithms were assessed. The risk of bias was evaluated using the quality assessment of diagnostic accuracy studies 2 (QUADAS-2) tool. A systematic search was conducted in EMBASE, Medline, and Scopus.

RESULTS

Out of 3410 identified records, 19 studies were included with six and seven studies having low risk of biases and applicability concerns for all the domains, respectively. Metrics varied widely and were assessed on multiple levels. F1-scores for classification and detection tasks were 68.3%-94.3% and 42.8%-95.4%, respectively. Irrespective of the task, F1-scores were 68.3%-95.4% for professional cameras, 78.8%-87.6%, for intraoral cameras, and 42.8%-80% for smartphone cameras. Limited studies allowed assessing AI performance for lesions of different severity.

CONCLUSION

Automatic detection of dental caries using AI may provide objective verification of clinicians' diagnoses and facilitate patient-clinician communication and teledentistry. Future studies should consider more robust study designs, employ comparable and standardized metrics, and focus on the severity of caries lesions.

Teeth Lesion Detection Using Deep Learning and the Internet of Things Post-COVID-19

With a view of the post-COVID-19 world and probable future pandemics, this paper presents an Internet of Things (IoT)-based automated healthcare diagnosis model that employs a mixed approach using data augmentation, transfer learning, and deep learning techniques and does not require physical interaction between the patient and physician. Through a user-friendly graphic user interface and availability of suitable computing power on smart devices, the embedded artificial intelligence allows the proposed model to be effectively used by a layperson without the need for a dental expert by indicating any issues with the teeth and subsequent treatment options. The proposed method involves multiple processes, including data acquisition using IoT devices, data preprocessing, deep learning-based feature extraction, and classification through an unsupervised neural network. The dataset contains multiple periapical X-rays of five different types of lesions obtained through an IoT device mounted within the mouth guard. A pretrained AlexNet, a fast GPU implementation of a convolutional neural network (CNN), is fine-tuned using data augmentation and transfer learning and employed to extract the suitable feature set. The data augmentation avoids overtraining, whereas accuracy is improved by transfer learning. Later, support vector machine (SVM) and the K-nearest neighbors (KNN) classifiers are trained for lesion classification. It was found that the proposed automated model based on the AlexNet extraction mechanism followed by the SVM classifier achieved an accuracy of 98%, showing the effectiveness of the presented approach.

Visual Diagnostics of Dental Caries through Deep Learning of Non-Standardised Photographs Using a Hybrid YOLO Ensemble and Transfer Learning Model

BACKGROUND

Access to oral healthcare is not uniform globally, particularly in rural areas with limited resources, which limits the potential of automated diagnostics and advanced tele-dentistry applications. The use of digital caries detection and progression monitoring through photographic communication, is influenced by multiple variables that are difficult to standardize in such settings. The objective of this study was to develop a novel and cost-effective virtual computer vision AI system to predict dental cavitations from non-standardised photographs with reasonable clinical accuracy.

METHODS

A set of 1703 augmented images was obtained from 233 de-identified teeth specimens. Images were acquired using a consumer smartphone, without any standardised apparatus applied. The study utilised state-of-the-art ensemble modeling, test-time augmentation, and transfer learning processes. The "you only look once" algorithm (YOLO) derivatives, v5s, v5m, v5l, and v5x, were independently evaluated, and an ensemble of the best results was augmented, and transfer learned with ResNet50, ResNet101, VGG16, AlexNet, and DenseNet. The outcomes were evaluated using precision, recall, and mean average precision (mAP).

RESULTS

The YOLO model ensemble achieved a mean average precision (mAP) of 0.732, an accuracy of 0.789, and a recall of 0.701. When transferred to VGG16, the final model demonstrated a diagnostic accuracy of 86.96%, precision of 0.89, and recall of 0.88. This surpassed all other base methods of object detection from free-hand non-standardised smartphone photographs.

CONCLUSION

A virtual computer vision AI system, blending a model ensemble, test-time augmentation, and transferred deep learning processes, was developed to predict dental cavitations from non-standardised photographs with reasonable clinical accuracy. This model can improve access to oral healthcare in rural areas with limited resources, and has the potential to aid in automated diagnostics and advanced tele-dentistry applications.

Advancements in Dentistry with Artificial Intelligence: Current Clinical Applications and Future Perspectives

Artificial intelligence has been widely used in the field of dentistry in recent years. The present study highlights current advances and limitations in integrating artificial intelligence, machine learning, and deep learning in subfields of dentistry including periodontology, endodontics, orthodontics, restorative dentistry, and oral pathology. This article aims to provide a systematic review of current clinical applications of artificial intelligence within different fields of dentistry. The preferred reporting items for systematic reviews (PRISMA) statement was used as a formal guideline for data collection. Data was obtained from research studies for 2009-2022. The analysis included a total of 55 papers from Google Scholar, IEEE, PubMed, and Scopus databases. Results show that artificial intelligence has the potential to improve dental care, disease diagnosis and prognosis, treatment planning, and risk assessment. Finally, this study highlights the limitations of the analyzed studies and provides future directions to improve dental care.

DCP: Prediction of Dental Caries Using Machine Learning in Personalized Medicine

Dental caries is an infectious disease that deteriorates the tooth structure, with tooth cavities as the most common result. Classified as one of the most prevalent oral health issues, research on dental caries has been carried out for early detection due to pain and cost of treatment. Medical research in oral healthcare has shown limitations such as considerable funds and time required; therefore, artificial intelligence has been used in recent years to develop models that can predict the risk of dental caries. The data used in our study were collected from a children’s oral health survey conducted in 2018 by the Korean Center for Disease Control and Prevention. Several Machine Learning algorithms were applied to this data, and their performances were evaluated using accuracy, F1-score, precision, and recall. Random forest has achieved the highest performance compared to other machine learnings methods, with an accuracy of 92%, F1-score of 90%, precision of 94%, and recall of 87%. The results of the proposed paper show that ML is highly recommended for dental professionals in assisting them in decision making for the early detection and treatment of dental caries.

A hybrid mask RCNN-based tool to localize dental cavities from real-time mixed photographic images

Nearly 3.5 billion humans have oral health issues, including dental caries, which requires dentist-patient exposure in oral examinations. The automated approaches identify and locate carious regions from dental images by localizing and processing either colored photographs or X-ray images taken via specialized dental photography cameras. The dentists' interpretation of carious regions is difficult since the detected regions are masked using solid coloring and limited to a particular dental image type. The software-based automated tools to localize caries from dental images taken via ordinary cameras requires further investigation. This research provided a mixed dataset of dental photographic (colored or X-ray) images, instantiated a deep learning approach to enhance the existing dental image carious regions' localization procedure, and implemented a full-fledged tool to present carious regions via simple dental images automatically. The instantiation mainly exploits the mixed dataset of dental images (colored photographs or X-rays) collected from multiple sources and pre-trained hybrid Mask RCNN to localize dental carious regions. The evaluations performed by the dentists showed that the correctness of annotated datasets is up to 96%, and the accuracy of the proposed system is between 78% and 92%. Moreover, the system achieved the overall satisfaction level of dentists above 80%.

Proof-of-Concept Study on an Automatic Computational System in Detecting and Classifying Occlusal Caries Lesions from Smartphone Color Images of Unrestored Extracted Teeth

Dental caries has been considered the heaviest worldwide oral health burden affecting a significant proportion of the population. To prevent dental caries, an appropriate and accurate early detection method is demanded. This proof-of-concept study aims to develop a two-stage computational system that can detect early occlusal caries from smartphone color images of unrestored extracted teeth according to modified International Caries Detection and Assessment System (ICDAS) criteria (3 classes: Code 0; Code 1-2; Code 3-6): in the first stage, carious lesion areas were identified and extracted from sound tooth regions. Then, five characteristic features of these areas were intendedly selected and calculated to be inputted into the classification stage, where five classifiers (Support Vector Machine, Random Forests, K-Nearest Neighbors, Gradient Boosted Tree, Logistic Regression) were evaluated to determine the best one among them. On a set of 587 smartphone images of extracted teeth, our system achieved accuracy, sensitivity, and specificity that were 87.39%, 89.88%, and 68.86% in the detection stage when compared to modified visual and image-based ICDAS criteria. For the classification stage, the Support Vector Machine model was recorded as the best model with accuracy, sensitivity, and specificity at 88.76%, 92.31%, and 85.21%. As the first step in developing the technology, our present findings confirm the feasibility of using smartphone color images to employ Artificial Intelligence algorithms in caries detection. To improve the performance of the proposed system, there is a need for further development in both in vitro and in vivo modeling. Besides that, an applicable system for accurately taking intra-oral images that can capture entire dental arches including the occlusal surfaces of premolars and molars also needs to be developed.

Recognizing Occlusal Caries in Dental Intraoral Images Using Deep Learning

Based on an image dataset of 88 in-vivo dental images taken with an intra-oral camera, we show that a Deep Learning model (Mask R-CNN) can detect and classify dental caries on occlusal surfaces across the whole 7-class ICDAS (International Caries Detection and Assessment System) scale. This is accomplished without any image pre-processing method and by utilizing superpixels segmentation for the experts' annotations and the evaluation of the classifier. In the proposed methodology, transfer learning and data augmentation are employed during the training of the model. The paper discusses technical details, provides initial results and denotes points for further improvement by fine-tuning the classifier along with an extended dataset.

Multi-Centre Clinical Evaluation of Photothermal Radiometry and Luminescence Correlated with International Benchmarks for Caries Detection

INTRODUCTION

A clinical study was initiated to investigate a caries detection device (The Canary System (CS)), based on photothermal radiometry and modulated luminescence (PTR-LUM). The primary objective of this study was to determine if PTR-LUM values (in the form of Canary Numbers; CN) correlate with International Caries Diagnostic and Assessment System (ICDAS II) scores and clinical situations. The secondary objectives of this study were to monitor the safety of PTR-LUM, and collect data to determine how CN values could be used to differentiate healthy from decayed tooth surfaces on a normalized scale.

METHODS

The trial was a four site, non-blinded study. Data was collected from 92 patients, resulting in 842 scanned tooth surfaces over multiple appointments. Surfaces were assessed according to ICDAS II, and further stratified into five clinical situation categories: 1) healthy surface, 2) non-cavitated white and/or brown spots; 3) caries lesions; 4) cavitation and 5) teeth undergoing remineralization therapy.CN data was analyzed separately for smooth and occlusal surfaces. Using a semi-logarithmic graph to plot raw CN (rCN) and normalized (CN) values, rCN data was normalized into a scale of 0-100.

RESULTS

Linear correlations (R2) between CN and ICDAS II groupings for smooth and occlusal surfaces were calculated as 0.9759 and 0.9267, respectively. The mean CN values derived from smooth (20.2±0.6) and occlusal (19±1.0) surfaces identified as healthy had significantly lower CN values (P<0.05) compared with the values from the other clinical situation categories. No adverse events were reported.

CONCLUSION

The present study demonstrated the safety of PTR-LUM for clinical application and its ability to distinguish sound from carious tooth surfaces. A clear shift from the baseline in both PTR and LUM in carious enamel was observed depending on the type and nature of the lesion, and correlated to ICDAS II classification codes, which enabled the preliminary development of a Canary Scale.

A computer-aided automated methodology for the detection and classification of occlusal caries from photographic color images

The aim of this work is to present a computer-aided automated methodology for the assessment of carious lesions, according to the International Caries Detection and Assessment System (ICDAS II), which are located on the occlusal surfaces of posterior permanent teeth from photographic color tooth images. The proposed methodology consists of two stages: (a) the detection of regions of interest and (b) the classification of the detected regions according to ICDAS ΙΙ. In the first stage, pre-processing, segmentation and post-processing mechanisms were employed. For each pixel of the detected regions, a 15×15 neighborhood is used and a set of intensity-based and texture-based features were extracted. A correlation based technique was applied to select a subset of 36 features which were given as input into the classification stage, where five classifiers (J48, Random Tree, Random Forests, Support Vector Machines and Naïve Bayes) were compared to conclude to the best one, in our case, to Random Forests. The methodology was evaluated on a set of 103 digital color images where 425 regions of interest from occlusal surfaces of extracted permanent teeth were manually segmented and classified, based on visual assessments by two experts. The methodology correctly detected 337 out of 340 regions in the detection stage with accuracy of detection 80%. For the classification stage an overall accuracy 83% is achieved. The proposed methodology provides an objective and fully automated caries diagnostic system for occlusal carious lesions with similar or better performance of a trained dentist taking into consideration the available medical knowledge.