Dental Images Research Articles

An intelligent deep network for dental medical image processing system

Nowadays, many people are affected by oral health issues because of continuous changes in lifestyle such as personal speech that is affected by crooked teeth and malocclusion teeth. Moreover, a dental problem can cause bacterial infections, cavities, and many other diseases due to an improper lifestyle. In this research, a novel Intelligent Ant Lion-based Convolution Neural Model (IALCNM) is designed for segmenting affected parts in teeth and to classify the wear and periodontitis diseases from the collected dataset. Moreover, the developed technique is implemented in the python 3.8 environment and the attained results of the developed procedure are related to other existing techniques implemented for different diseases in standings of accuracy, precision, error rate, execution time, and so on. Hence the outcome indicates that the current research technique applied to self-created datasets has enhanced the accuracy of segmenting affected parts and disease prediction more than other techniques.

A survey of patient contact shielding in dental teaching hospitals in the UK

A high volume of dental imaging is carried out each year. In the UK, guidance on the use of patient contact shielding for these investigations is published by the British Institute of Radiology (BIR) and in a document jointly produced by the Faculty of General Dental Practice and Public Health England (FGDP/PHE). Both these sources of guidance have been updated recently and patient contact shielding is no longer recommended for most imaging settings in dental radiology. It is unclear whether radiology departments at dental teaching hospitals in the UK are aware of these sources of guidance, and how this relates to clinical practice within these departments. A survey was carried out exploring the awareness of current guidelines and clinical practice at dental teaching hospitals in the UK. The survey was sent to a representative at 17 different dental teaching hospital radiology departments. Responses were received from 11 departments. The range of intra-oral and extra-oral imaging carried out at these departments was comparable. Ten departments were aware of the existence of national guidelines for patient contact shielding, however only four were specifically aware of the recent BIR guidelines and only four were specifically aware of the FGDP/PHE guidelines. No department was aware of both sets of guidelines. No departments used thyroid protection for bitewing, periapical, lower 45 degree occlusal, panoramic or lateral cephalometric radiographs. Six departments sometimes or always used thyroid protection for upper standard occlusal imaging. Two departments used thyroid protection for cone beam CT imaging. No departments routinely used lead aprons on patients for dental imaging. In conclusion, radiology departments at dental teaching hospitals in the UK do not use patient contact shielding for most imaging situations in dental radiology. There is mixed awareness of current national guidelines, but the reported clinical practice aligns well with the current guidelines.

Automatic Segmentation of Periapical Radiograph Using Color Histogram and Machine Learning for Osteoporosis Detection.

Osteoporosis leads to the loss of cortical thickness, a decrease in bone mineral density (BMD), deterioration in the size of trabeculae, and an increased risk of fractures. Changes in trabecular bone due to osteoporosis can be observed on periapical radiographs, which are widely used in dental practice. This study proposes an automatic trabecular bone segmentation method for detecting osteoporosis using a color histogram and machine learning (ML), based on 120 regions of interest (ROI) on periapical radiographs, and divided into 60 training and 42 testing datasets. The diagnosis of osteoporosis is based on BMD as evaluated by dual X-ray absorptiometry. The proposed method comprises five stages: the obtaining of ROI images, conversion to grayscale, color histogram segmentation, extraction of pixel distribution, and performance evaluation of the ML classifier. For trabecular bone segmentation, we compare K-means and Fuzzy C-means. The distribution of pixels obtained from the K-means and Fuzzy C-means segmentation was used to detect osteoporosis using three ML methods: decision tree, naive Bayes, and multilayer perceptron. The testing dataset was used to obtain the results in this study. Based on the performance evaluation of the K-means and Fuzzy C-means segmentation methods combined with 3 ML, the osteoporosis detection method with the best diagnostic performance was K-means segmentation combined with a multilayer perceptron classifier, with accuracy, specificity, and sensitivity of 90.48%, 90.90%, and 90.00%, respectively. The high accuracy of this study indicates that the proposed method provides a significant contribution to the detection of osteoporosis in the field of medical and dental image analysis.

Exploiting multimodal CNN architecture for automated teeth segmentation on dental panoramic X-ray images.

Panoramic X-ray images are the major source used in field of dental image segmentation. However, such images suffers from the disturbances like low contrast, presence of jaw bones, nose bones, spinal bone, and artifacts. Thus, to observe these images manually is a tedious task, requires expertise of dentist and is time consuming. Hence, there is need to develop an automated tool for teeth segmentation. Recently, few deep models have been developed for dental image segmentation. But, such models possess large number of training parameters, thus making the segmentation a very complex task. Also, these models are based only on conventional CNN and lacks in exploiting multimodal CNN features for dental image segmentation. Thus, to address these issues, a novel encoder-decoder model based on multimodal-feature extraction for automatic segmentation of teeth area is proposed. The encoder has three different CNN based architectures: conventional CNN, atrous-CNN, and separable CNN to encode rich contextual information. Whereas decoder contains a single stream of deconvolutional layers for segmentation. The proposed model is tested on 1500 panoramic X-ray images and uses very less parameters when compared to state-of-the-art methods. Besides this, the precision and recall are 95.01% and 94.06%, which out performs the state-of-the art methods.

Application of liquid scintillation light guide (LSLG) to scattered X-ray measurement from dental panoramic radiography system

A 1.2 m flexible liquid scintillation light guide (LSLG) detector connected to a portable electric device was developed and applied to monitor X-rays scattered from a panoramic dental X-ray imaging apparatus. The X-ray absorption of the LSLG tube was simulated for 20 and 40 keV X-rays. The LSLG detector was calibrated by comparing doses measured by a 3-inch NaI(Tl) detector using Am-241 and Cs-137 sources. A linear relationship was obtained between dose rates (μSv/h) and count rates (cps). The elapsed time profile of scattered radiation from a panoramic dental X-ray imaging apparatus was determined. Local absorbed doses were measured with small OSL dosimeters on the LSLG tube suspended over the shoulder and the eyeglass. Behind the ears and neck (in the center of the LSLG tube) showed high doses. The LSLG detector was useful for real-time monitoring of scattered X-rays.

Quantitative level determination of fixed restorations on panoramic radiographs using deep learning.

Although many studies in various fields employ deep learning models, only few such studies exist in dental imaging. This paper aims to evaluate the effectiveness of Convolutional Neural Network (CNN) algorithms for the detection and diagnosis of the quantitative level of dental restorations using panoramic radiographs by preparing a novel dataset. 20,973 panoramic X-ray radiographs, all labeled into five distinct categories by three dental experts, were used. AlexNet, VGG-16 and variants of ResNet models were trained with the dataset and evaluated for the classification task. Additionally, 10-fold cross-validation (i.e. nine-folds were separated for training and one-fold for validation) and data augmentation were carried out for all experiments. The most successful result came from ResNet-101 with 92.7% accuracy. Its macro-average AUC was also the highest with 0.989. Other accuracy results obtained for the dataset were 75.5% for AlexNet, 85.0% for VGG-16, 92.1% for ResNet-18, 91.7% for ResNet-50 and 92.1% for Inception ResNet V2. Accuracy of 92.7% is a very promising result for a computer-aided diagnosis system. This result proved that the system could assist dentists in providing supportive preliminary information the moment they first take a panoramic X-ray radiograph of a patient. Furthermore, as the introduced dataset is powerful enough, it can be re-labeled for different problems and used in different studies.

Deep Learning-Based Multiclass Instance Segmentation for Dental Lesion Detection.

Automated dental imaging interpretation is one of the most prolific areas of research using artificial intelligence. X-ray imaging systems have enabled dental clinicians to identify dental diseases. However, the manual process of dental disease assessment is tedious and error-prone when diagnosed by inexperienced dentists. Thus, researchers have employed different advanced computer vision techniques, as well as machine and deep learning models for dental disease diagnoses using X-ray imagery. In this regard, a lightweight Mask-RCNN model is proposed for periapical disease detection. The proposed model is constructed in two parts: a lightweight modified MobileNet-v2 backbone and region-based network (RPN) are proposed for periapical disease localization on a small dataset. To measure the effectiveness of the proposed model, the lightweight Mask-RCNN is evaluated on a custom annotated dataset comprising images of five different types of periapical lesions. The results reveal that the model can detect and localize periapical lesions with an overall accuracy of 94%, a mean average precision of 85%, and a mean insection over a union of 71.0%. The proposed model improves the detection, classification, and localization accuracy significantly using a smaller number of images compared to existing methods and outperforms state-of-the-art approaches.

Metal Artifact Reduction in Dental CBCT Images Using Direct Sinogram Correction Combined with Metal Path-Length Weighting.

Metal artifacts in dental computed tomography (CT) images, caused by highly X-ray absorbing objects, such as dental implants or crowns, often more severely compromise image readability than in medical CT images. Since lower tube voltages are used for dental CTs in spite of the more frequent presence of metallic objects in the patient, metal artifacts appear more severely in dental CT images, and the artifacts often persist even after metal artifact correction. The direct sinogram correction (DSC) method, which directly corrects the sinogram using the mapping function derived by minimizing the sinogram inconsistency, works well in the case of mild metal artifacts, but it often fails to correct severe metal artifacts. We propose a modified DSC method to reduce severe metal artifacts, and we have tested it on human dental images. We first segment the metallic objects in the CT image, and then we forward-project the segmented metal mask to identify the metal traces in the projection data with computing the metal path length for the rays penetrating the metal mask. In the sinogram correction with the DSC mapping function, we apply the weighting proportional to the metal path length. We have applied the proposed method to the phantom and patient images taken at the X-ray tube voltage of 90 kVp. We observed that the proposed method outperforms the original DSC method when metal artifacts were severe. However, we need further extensive studies to verify the proposed method for various CT scan conditions with many more patient images.

Blind denoising of dental X-ray images

The paper considers a problem of automatic analysis and noise suppression in dental X-Ray images, e.g., in images acquired by dental Morita system. Such images contain spatially correlated noise with unknown spectrum and with standard deviation that varies for different image regions. In the paper, we propose two deep convolutional neural networks. The first network estimates the spectrum and level of noise for each pixel of a noisy image, predicting maps of noise standard deviation for three image scales. The second network uses the maps as inputs to suppress noise in the image. It is shown, using modelled and real-life images, that the proposed networks provide PSNR for dental X-Ray images by 2.7 dB better than other modern denoising methods.

Reliability and accuracy of intraoral radiography, cone beam CT, and dental MRI for evaluation of peri-implant bone lesions at zirconia implants − an ex vivo feasibility study

ObjectivesTo determine the reliability and accuracy of intraoral radiography (IR), cone-beam-computed tomography (CBCT), and dental magnetic resonance imaging (dMRI) in measuring peri‑implant bone defects around single zirconia implants. MethodsTwenty-four zirconia implants were inserted in bovine ribs with various peri‑implant defect sizes and morphologies. True defect extent was measured without implant in CBCT. Defects were measured twice in IR, CBCT, and dMRI with the inserted implant by three experienced readers. Reliability was assessed by ICC, accuracy by the Friedman test, and post-hoc-Tukey's test. ResultsA comparable good to excellent intra- and inter-reader reliability was observed for all modalities (intra-/inter-rater-CC range for IR; CBCT; dMRI: 0.81–0.91/0.79;0.87–0.97/0.96;0.87–0.95/0.94). Accuracy was generally high, with mean errors below 1 mm in all directions. However, measuring defect depth in the mesiodistal direction was significantly more accurate in dMRI (0.65 ± 0.38 mm) compared to IR (2.71 ± 1.91 mm), and CBCT (1.98 ± 1.97 mm), p-values ≤ 0.0001 respectively ≤ 0.01. ConclusionsOsseous defects around zirconia implants can be reliably measured in IR/CBCT/dMRI in the mesiodistal directions. In addition, CBCT and dMRI allow assessment of the buccolingual directions. dMRI provides a comparable accuracy in all directions, except for the mesiodistal defect depth, where it outperforms IR and CBCT.

Dental caries detection using a semi-supervised learning approach

Early diagnosis of dental caries progression can prevent invasive treatment and enable preventive treatment. In this regard, dental radiography is a widely used tool to capture dental visuals that are used for the detection and diagnosis of caries. Different deep learning (DL) techniques have been used to automatically analyse dental images for caries detection. However, most of these techniques require large-scale annotated data to train DL models. On the other hand, in clinical settings, such medical images are scarcely available and annotations are costly and time-consuming. To this end, we present an efficient self-training-based method for caries detection and segmentation that leverages a small set of labelled images for training the teacher model and a large collection of unlabelled images for training the student model. We also propose to use centroid cropped images of the caries region and different augmentation techniques for the training of self-supervised models that provide computational and performance gains as compared to fully supervised learning and standard self-supervised learning methods. We present a fully labelled dental radiographic dataset of 141 images that are used for the evaluation of baseline and proposed models. Our proposed self-supervised learning strategy has provided performance improvement of approximately 6% and 3% in terms of average pixel accuracy and mean intersection over union, respectively as compared to standard self-supervised learning. Data and code will be made available to facilitate future research.

Fiber Optic Plate Coupled Pb-Free Perovskite X-ray Camera Featuring Low-Dose-Rate Imaging toward Dental Diagnosis.

Copper-based halide perovskites have been considered as promising scintillators. However, they still cannot meet the requirement of low-dose-rate X-ray imaging in medical diagnosis. Herein, we design a fiber optic plate (FOP) coupled perovskite X-ray camera to reduce the dose rate toward dental X-ray imaging. Tl doped Cs3Cu2I5 prepared via molten salt reaction has a high light yield of 72,000 photons/MeV, resulting from Tl10/Tl20-self-trapped hole emissions. After FOP coupling, the pulp cavity, root canal, dentin and root canal file can be clearly observed under a low dose rate as low as 3 μGyair s-1, which is absolutely lower than the required 5.5 μGyair s-1 for commercial intraoral dental sensors. The realization of such a low dose rate is attributed to the high coupling efficiency of 75% for the FOP and the high brightness of 262 lm m-2 for the scintillation screen. This designed portable X-ray camera shows its huge potential in intraoral dental X-ray imaging.

IMAGING IN DENTISTRY

Introduction: Since the discovery of the x-ray, imaging has progressed a lot and is applied in various areas of medicine and dentistry, the latter being an area that has undergone great evolution with the introduction of radiography, which came to complement and solve doubts about the exam and allow the dentist to make a more precise assessment and conduct. Objective: To highlight the two-dimensional and three-dimensional radiographic techniques used in dentistry, emphasizing their indications, their advantages and disadvantages, as well as their relevant importance in dental practice. Methodology: This is a narrative review of the scientific literature in which data were collected through electronic searches in different databases using the descriptors Imagiology, Dentistry and Dental Imaging with the Boolean operator “AND”. Publications in Portuguese and English from the last 20 (twenty) years whose main focus was imaging applied to dentistry were used. Results and Discussion: When it comes to two-dimensional techniques, the three types of intraoral radiographs are periapical radiography, interproximal radiography and occlusal radiography. Three-dimensional techniques include computed tomography and cone-beam computed tomography. Each of these techniques has its indications, particularities, advantages and disadvantages that must always be evaluated and indicated according to the need. Conclusion: There is no doubt about the benefits that the radiographic techniques commonly used in dental practice introduce in the diagnosis, planning and treatment and in the different areas of work of the dental surgeon. It is always up to the professional to act individually with patients and evaluate the best technique to be adopted for each situation.

Research Progress in the Application of Terahertz Spectroscopy and Imaging Technology in Stomatology

Terahertz waves, the electromagnetic waves in the range of 0.1 to 10 THz, has the advantages of being damage-free, causing no ionizing radiation injury, and being capable of recognizing the fingerprint spectrum of molecular characteristics, thus holding encouraging prospects for wide applications in the field of biomedicine. Terahertz spectrum can be used to identify and characterize biological structures of different levels, from biomolecules such as proteins to cells and tissues, through the spectral signals and/or restored images of the samples. Herein, we summarized the current stomatogical application of and research progress in terahertz spectroscopy and imaging in dentistry, reported the latest research findings, strengths and limitations from three perspectives, tooth anatomical structure, the extent of caries progression, and oral soft tissue, and suggested possible directions for future exploration.

Prevalence of missed canals in endodontically treated teeth: A cone-beam computed tomography study.

This study aimed to measure the prevalence of untreated root canals and periapical lesions in endodontically treated teeth using cone-beam computed tomography (CBCT) and their association with the coronal seal. This cross-sectional study assessed CBCT images of studied patients alongside documented information from their dental examinations. The assessment method for analyzing CBCT images after including endodontically treated teeth was the presence of missed canals, and periapical lesions were analyzed in three dimensions. The evaluations from 772 teeth demonstrated that 89.4% of teeth had the proper coronal seal, 13.3% owned missed root canals, and 29.4% had periapical lesions. Most untreated canals belonged to maxillary first molars (71.8%) and mandibular lateral incisors (33.3%). The prevalence of periapical lesions was highest in mandibular lateral incisors (58.3%) and maxillary second molars (55.8%). In teeth with missed canals, the most periapical lesions were observed in the first premolars of both jaws. In teeth with the lack of proper coronal seal, the periapical lesions were observed in 55.6% of teeth with untreated root canals. The results revealed a high prevalence of missed canals and periapical lesions in endodontically treated teeth in our study population. Key words:Untreated Root Canals, Periapical Lesions, 3-D Dental Imaging, Dental Treatment Failure.

Enhancing CNN for Forensics Age Estimation Using CGAN and Pseudo-Labelling

Age estimation using forensics odontology is an important process in identifying victims in criminal or mass disaster cases. Traditionally, this process is done manually by human expert. However, the speed and accuracy may vary depending on the expertise level of the human expert and other human factors such as level of fatigue and attentiveness. To improve the recognition speed and consistency, researchers have proposed automated age estimation using deep learning techniques such as Convolutional Neural Network (CNN). CNN requires many training images to obtain high percentage of recognition accuracy. Unfortunately, it is very difficult to get large number of samples of dental images for training the CNN due to the need to comply to privacy acts. A promising solution to this problem is a technique called Generative Adversarial Network (GAN). GAN is a technique that can generate synthetic images that has similar statistics as the training set. A variation of GAN called Conditional GAN (CGAN) enables the generation of the synthetic images to be controlled more precisely such that only the specified type of images will be generated. This paper proposes a CGAN for generating new dental images to increase the number of images available for training a CNN model to perform age estimation. We also propose a pseudo-labelling technique to label the generated images with proper age and gender. We used the combination of real and generated images to train Dental Age and Sex Net (DASNET), which is a CNN model for dental age estimation. Based on the experiment conducted, the accuracy, coefficient of determination (R2) and Absolute Error (AE) of DASNET have improved to 87%, 0.85 and 1.18 years respectively as opposed to 74%, 0.72 and 3.45 years when DASNET is trained using real, but smaller number of images.

CVIP-Net: A Convolutional Neural Network-Based Model for Forensic Radiology Image Classification

Automated and autonomous decisions of image classification systems have essential applicability in this modern age even. Image-based decisions are commonly taken through explicit or auto-feature engineering of images. In forensic radiology, auto decisions based on images significantly affect the automation of various tasks. This study aims to assist forensic radiology in its biological profile estimation when only bones are left. A benchmarked dataset Radiology Society of North America (RSNA) has been used for research and experiments. Additionally, a locally developed dataset has also been used for research and experiments to cross-validate the results. A Convolutional Neural Network (CNN)-based model named computer vision and image processing-net (CVIP-Net) has been proposed to learn and classify image features. Experiments have also been performed on state-of-the-art pertained models, which are alex_net, inceptionv_3, google_net, Residual Network (resnet)_50, and Visual Geometry Group (VGG)-19. Experiments proved that the proposed CNN model is more accurate than other models when panoramic dental x-ray images are used to identify age and gender. The specially designed CNN-based achieved results in terms of standard evaluation measures including accuracy (98.90%), specificity (97.99%), sensitivity (99.34%), and Area under the Curve (AUC)-value (0.99) on the locally developed dataset to detect age. The classification rates of the proposed model for gender estimation were 99.57%, 97.67%, 98.99%, and 0.98, achieved in terms of accuracy, specificity, sensitivity, and AUC-value, respectively, on the local dataset. The classification rates of the proposed model for age estimation were 96.80%, 96.80%, 97.03%, and 0.99 achieved in terms of accuracy, specificity, sensitivity, and AUC-value, respectively, on the RSNA dataset.

3D Bitemark Analysis in Forensic Odontology Utilizing a Smartphone Camera and Open-Source Monoscopic Photogrammetry Surface Scanning

Bitemark analysis is a challenging procedure in the field of criminal case investigation. The unique characteristics of dentition are used to find the best match between the existing patterned injury and the suspected perpetrator in bitemark identification. Bitemark analysis accuracy can be influenced by various factors, including biting pressure, tooth morphology, skin elasticity, dental cast duplication, timing, and image quality. This review article discusses the potential of a smartphone camera as an alternative method for 3D bitemark analysis. Bitemark evidence on human skin and food should be immediately recorded or duplicated to retrieve long-lasting proof, allowing for a sufficient examination period. Various studies utilizing two-dimensional (2D) and three-dimensional (3D) technologies have been developed to obtain an adequate bitemark analysis. 3D imaging technology provides accurate and precise analysis. However, the currently available method using an intraoral scanner (IOS) requires high-cost specialized equipment and a well-trained operator. The numerous advantages of monoscopic photogrammetry may lead to a novel method of 3D bitemark analysis in forensic odontology. Smartphone cameras and monoscopic photogrammetry methodology could lead to a novel method of 3D bitemark analysis with an efficient cost and readily available equipment.

A pilot study of half-value layer measurements using a semiconductor dosimeter for intraoral radiography.

This pilot study was conducted to evaluate half-value layer (HVL) measurements obtained using a semiconductor dosimeter for intraoral radiography. This study included 8 aluminum plates, 4 of which were low-purity (less than 99.9%) and 4 high-purity (greater than 99.9%). Intraoral radiography was performed using an intraoral X-ray unit in accordance with the dental protocol at the authors' affiliated hospital: tube voltage, 60 kVp and 70 kVp; tube current, 7 mA; and exposure time, 0.10 s. The accuracy of HVL measurements for intraoral radiography was assessed using a semiconductor dosimeter. A simple regression analysis was performed to compare the aluminum plate thickness and HVL in relation to the tube voltage (60 kVp and 70 kVp) and aluminum purity (low and high). For the low-purity aluminum plates, the HVL at 60 kVp (Y) and 70 kVp (Y) was significantly correlated with the thickness of the aluminum plate (X), with Y = 1.708 + 0.415X (r=0.999, P<0.05) and Y = 1.980 + 0.484X (r=0.999, P<0.05), respectively. Similarly, for the high-purity aluminum plates, the HVL at 60 kVp (Y) and 70 kVp (Y) was significantly correlated with the plate thickness (X), with Y = 1.696 + 0.454X (r=0.999, P<0.05) and Y = 1.968 + 0.515X (r=0.998, P<0.05), respectively. This pilot study examined the relationship between aluminum plate thickness and HVL measurements using a semiconductor dosimeter for intraoral radiography. Semiconductor dosimeters may prove useful in HVL measurement for purposes such as quality assurance in dental X-ray imaging.

Deep learning-based gender classification with dental X-ray images

Deep learning-based gender classification with dental X-ray images