Enhancing Caries Detection in Bitewing Radiographs Using YOLOv7.

The study aimed to evaluate the impact of image size, area of detection (IoU) thresholds and confidence thresholds on the performance of the YOLO models in the detection of dental caries in bitewing radiographs. A total of 2575 bitewing radiographs were annotated with seven classes according to the ICCMS™ radiographic scoring system. YOLOv3 and YOLOv7 models were employed with different configurations, and their performances were evaluated based on precision, recall, F1-score and mean average precision (mAP). Results showed that YOLOv7 with 640 × 640 pixel images exhibited significantly superior performance compared to YOLOv3 in terms of precision (0.557 vs. 0.268), F1-score (0.555 vs. 0.375) and mAP (0.562 vs. 0.458), while the recall was significantly lower (0.552 vs. 0.697). The following experiment found that the overall mAPs did not significantly differ between 640 × 640 pixel and 1280 × 1280 pixel images, for YOLOv7 with an IoU of 50% and a confidence threshold of 0.001 (p = 0.866). The last experiment revealed that the precision significantly increased from 0.570 to 0.593 for YOLOv7 with an IoU of 75% and a confidence threshold of 0.5, but the mean-recall significantly decreased and led to lower mAPs in both IoUs. In conclusion, YOLOv7 outperformed YOLOv3 in caries detection and increasing the image size did not enhance the model's performance. Elevating the IoU from 50% to 75% and confidence threshold from 0.001 to 0.5 led to a reduction of the model's performance, while simultaneously improving precision and reducing recall (minimizing false positives and negatives) for carious lesion detection in bitewing radiographs.

Reproducibility analysis of automated deep learning based localisation of mandibular canals on a temporal CBCT dataset.

Preoperative radiological identification of mandibular canals is essential for maxillofacial surgery. This study demonstrates the reproducibility of a deep learning system (DLS) by evaluating its localisation performance on 165 heterogeneous cone beam computed tomography (CBCT) scans from 72 patients in comparison to an experienced radiologist's annotations. We evaluated the performance of the DLS using the symmetric mean curve distance (SMCD), the average symmetric surface distance (ASSD), and the Dice similarity coefficient (DSC). The reproducibility of the SMCD was assessed using the within-subject coefficient of repeatability (RC). Three other experts rated the diagnostic validity twice using a 0-4 Likert scale. The reproducibility of the Likert scoring was assessed using the repeatability measure (RM). The RC of SMCD was 0.969 mm, the median (interquartile range) SMCD and ASSD were 0.643 (0.186) mm and 0.351 (0.135) mm, respectively, and the mean (standard deviation) DSC was 0.548 (0.138). The DLS performance was most affected by postoperative changes. The RM of the Likert scoring was 0.923 for the radiologist and 0.877 for the DLS. The mean (standard deviation) Likert score was 3.94 (0.27) for the radiologist and 3.84 (0.65) for the DLS. The DLS demonstrated proficient qualitative and quantitative reproducibility, temporal generalisability, and clinical validity.

Feasibility of deep learning for dental caries classification in bitewing radiographs based on the ICCMS™ radiographic scoring system.

OBJECTIVE: To evaluate the potential of deep learning models for categorization of dental caries in bitewing radiographs based on the International Caries Classification and Management System (ICCMS™) radiographic scoring system (RSS). STUDY DESIGN: In total, 2758 annotated bitewing radiographs were randomly divided into 3 experiments to assess the ResNet-18, -50, -101, and -152. Experiment A tested 4-class ICCMS™-RSS training and validation using Carestream (CS) radiographs; experiment B tested training and validation using CS and VistaScan radiographs; experiment C tested 7-class ICCMS™-RSS training and validation using CS and VistaScan radiographs. The performance matrices and the areas under the receiver operating characteristic curves were analyzed to assess all procedures. RESULTS: In experiment A, ResNet-50 and ResNet-152 were equally accurate (71.11%) and approximately 78% sensitive. The latter presented the highest specificity (56.90%). In experiment B, ResNet-50 presented the highest sensitivity (79.51%) but ResNet-152 had the highest specificity (60.71%). In experiment C, all models markedly underperformed in distinguishing the 7-class ICCMS™-RSS with specificities of 16.46% to 22.41%. They had fewer classification errors in the 4-class classification (28.89%-35.56%) than in the 7-class classification (42.34%-53.06%). The areas under the receiver operating characteristic curves of all models were unanimously comparable. CONCLUSIONS: The ResNet models were able to classify dental caries according to the ICCMS™-RSS with average performances. The models underperformed in complicated classification tasks.

Assessment of YOLOv3 for caries detection in bitewing radiographs based on the ICCMS™ radiographic scoring system.

OBJECTIVES: To assess the feasibility of the YOLOv3 model under the intersection over union (IoU) thresholds of 0.5 (IoU50) and 0.75 (IoU75) for caries detection in bitewing radiographs based on the International Caries Classification and Management System (ICCMS™). MATERIALS AND METHODS: We trained the YOLOv3 model by feeding 994 annotated radiographs with the IoU50 and IoU75 thresholds. The testing procedure (n = 175) was subsequently conducted to evaluate the model's prediction metrics on caries classification based on the ICCMS™ radiographic scoring system. RESULTS: Regarding the 4-class classification representing caries severity, YOLOv3 could accurately detect and classify enamel caries and initial dentin caries (class RA) (IoU50 vs IoU75: precision, 0.75 vs 0.71; recall, 0.67 vs 0.64). Concerning the 7-class classification signifying specific caries depth (class 0, healthy tooth; classes RA1-3, initial caries affecting outer half, inner half of enamel, and the outer 1/3 of dentin; class RB4, caries extending to the middle 1/3 of dentin; classes RC5-6, extensively cavitated caries affecting the inner 1/3 of dentin and involving the pulp chamber), YOLOv3 could accurately detect and classify caries with pulpal exposure (class RC6) (IoU50 vs IoU75: precision, 0.77 vs 0.73; recall, 0.61 vs 0.57) but it failed to predict the outer half of enamel caries (class RA1) (IoU50 vs IoU75: precision, 0.35 vs 0.32; recall, 0.23 vs 0.21). CONCLUSIONS: YOLOv3 yielded acceptable performances in both IoU50 and IoU75. Although the performance metrics decreased in the 7-class detection, the two thresholds revealed comparable results. However, the model could not consistently detect initial-stage caries affecting the outermost surface of the enamel. CLINICAL RELEVANCE: YOLOv3 could be implemented to detect and classify dental caries according to the ICCMS™ classification with acceptable performances to assist dentists in making treatment decisions.

Effects of the cone-beam computed tomography protocol on the accuracy and image quality of root surface area measurements: An in vitro study.

Purpose: The objective of this study was to evaluate and compare the accuracy and image quality of root surface area (RSA) measurements obtained with various cone-beam computed tomography (CBCT) protocols, relative to the gold standard of micro-computed tomography (CT), in an in vitro setting. Materials and Methods: Four dry human skulls were scanned using 8 different protocols, with voxel sizes of 0.15 mm, 0.3 mm, and 0.4 mm. Three-dimensional models of the selected teeth were constructed using CBCT and micro-CT protocols, and the RSA was automatically measured by the image-processing software. The absolute difference in the percentage of the RSA (%ΔRSA) was calculated and compared across the 8 CBCT protocols using repeated-measures analysis of variance. Finally, image quality scores of the RSA measurements were computed and reported in terms of percent distribution. Results: No significant differences were observed in the %ΔRSA across the 8 protocols (P>0.05). The deviation in %ΔRSA ranged from 1.51% to 4.30%, with an increase corresponding to voxel size. As the voxel size increased, the image quality deteriorated. This decline in quality was particularly noticeable at the apical level of the root, where the distribution of poorer scores was most concentrated. Conclusion: Relative to CBCT protocols with voxel sizes of 0.15 mm and 0.3 mm, the protocols with a voxel size of 0.4 mm demonstrated inferior image quality at the apical levels. In spite of this, no significant discrepancies were observed in RSA measurements across the different CBCT protocols.

The development and testing of Thai facial soft tissue thickness data in three-dimensional computerized forensic facial reconstruction.

Forensic facial reconstruction is a useful tool to assist the public in recognizing human remains, leading to positive forensic investigation outcomes. To reproduce a virtual face, facial soft tissue thickness is one of the major guidelines to reach the accuracy and reliability for three-dimensional computerized facial reconstruction, a method that is making a significant contribution to improving forensic investigation and identification. This study aimed to develop a facial soft tissue thickness dataset for a Thai population, and test its reliability in the context of facial reconstruction. Three-dimensional facial reconstruction was conducted on four skulls (2 males and 2 females, with ages ranging between 51 to 60 years). Two main tools of three-dimensional computer animation and modeling software-Blender and Autodesk Maya-were used to rebuild the three-dimensional virtual face. The three-dimensional coordinate (x, y, z) cutaneous landmarks on the mesh templates were aligned homologous to the facial soft tissue thickness markers on the three-dimensional skull model. The final three-dimensional virtual face was compared to the target frontal photograph using face pool comparison. Four three-dimensional virtual faces were matched at low to moderate levels, ranging from 30% to 70% accuracy. These results demonstrate that the facial soft tissue thickness database of a Thai population applied in this study could be useful for three-dimensional computerized facial reconstruction purposes.

Comparing the pulp/tooth area ratio and dentin thickness of mandibular first molars in different age groups: A cone-beam computed tomography study.

CONTEXT: Mandibular first molar frequently requires endodontic treatment. Understanding age-related changes in pulp-dentin complex and root canal morphologies is essential for successful endodontic and restorative treatments. AIM: This study aimed to compare pulp/tooth area ratio (PTAR) and dentin thickness (DT) in mandibular first molars in different age groups through cone-beam computed tomography (CBCT) imaging. SUBJECTS AND METHODS: One hundred CBCT images of mandibular first molar were divided into five groups; age 20-29, 30-39, 40-49, 50-59, and 60 years old and older. Axial images were used to determine PTAR at Level A (furcation), Level B (between Levels A and C), and Level C (half distance between the furcation and apex of the root). The minimum DT of the distal wall of mesiobuccal (MB) and mesiolingual (MLi) canal and mesial wall of distal canal at 2 and 3 mm under the furcation was measured. STATISTICAL ANALYSIS USED: Analysis of variance was used to determine differences among age groups. RESULTS: PTAR was determined to reduce as age increases, showing a significant difference among the age groups at Levels A, B, and C of both roots (P < 0.05). The minimum DT was found to increase with age, demonstrating a significant difference among the age groups of MB and MLi canal at 2 and 3 mm (P < 0.05). No statistically significant difference was observed in the mesial DT of distal canal. CONCLUSIONS: The reduction of PTAR and the increasing DT were confirmed with advanced age.

Three-Dimensional Prediction of the Nose for Facial Approximation in a Thai Population.

Most of the previous studies about nose prediction were concentrated only some landmarks of the nose. This study aimed to generate prediction equations for ten landmarks of the nose in the midline and alar regions for forensic facial approximation. The six midline landmarks were the sellion, nasion-pronasale posterior, nasion-pronasale anterior, pronasale, nasal drop, and subnasale. The four landmarks of the alar region were the alare, superior alar groove, posterior alar groove, and inferior alar groove. We used the skull landmarks in the nasal, zygomatic, and maxillary bone to predict the landmarks of the nose. Cone-beam computed tomography scans of 108 Thai subjects with normal BMI and age ranging from 21.0 to 50.9 years were obtained in a sitting position. The data were converted into three-dimensional (3-D) images of the skull and face. The Cartesian coordinates of the landmarks of the skull and nose were used to formulate the multiple regression equations. The formulated equations were tested in 24 new subjects. The mean differences in the predicted midline landmarks varied between -0.4 mm and 0.5 mm, whereas those for bilateral landmarks varied between -1.0 mm and 1.4 mm. In conclusion, the prediction equations formulated here will be beneficial for facial approximation of the nose in a Thai population.

Prevalence of pre-eruptive intracoronal radiolucencies in Thai children and adolescents: A retrospective study.

AIM: The aim of the present study was to determine the prevalence, location, size, and tooth type of pre-eruptive intracoronal radiolucencies (PEIR) from panoramic radiographs in Thai patients. METHODS: A retrospective study of PEIR was conducted on 1599 panoramic radiographs of 629 Thai males and 970 Thai females with an age range of 4-20 years (average age: 14.0 ± 4.0 years). Each radiograph was reviewed by two calibrated examiners. The age and sex of the patients, location, size, and tooth type of the lesions were recorded. The SPSS program was used for statistical analysis. RESULTS: The PEIR were found in 0.32% of tooth samples (29/9060) and 1.63% of patients (26/1599). There was no statistical difference in the presence of PEIR between both sexes. The mandibular second molars were the most affected teeth. The lesions were mostly located at the central part of the crown. The lesion sizes were mostly limited to less than one-third of dentinal thickness. CONCLUSIONS: The present study showed that the prevalence of PEIR in Thai samples is 1.63%. This could raise dentists' awareness in reviewing unerupted teeth in panoramic radiographs. Early detection of the lesions should aid in proper treatment planning of affected teeth to prevent complications from carious invasion after tooth eruption.

Infection control and patient discomfort with an alternative plastic barrier in intraoral digital radiography.

OBJECTIVES: (1) To compare the efficacy of a commercially available hygienic sheath and an alternative plastic bag in preventing contamination of the imaging plate during intraoral radiography and (2) to compare patient discomfort when using the hygienic sheath and the plastic bag. METHODS: 60 sterilized Size 2 imaging plates covered with either the hygienic sheath (n = 30) or the plastic bag (n = 30) were used to simulate digital periapical radiographic examination in 30 volunteer patients. After disinfection, each plate was swabbed. The swabbed medium was then plated on trypticase soy agar and incubated. Bacterial colonies were counted. Patient discomfort was assessed using a visual analogue scale (VAS) score. The comparison of the number of bacterial colonies and VAS scores between the two groups was tested by paired t-test at p < 0.05. RESULTS: There was no significant difference in the number of bacterial colonies between the two groups (p = 0.745). Of all the plates, 10% plates yielded bacterial colonies. The mean count of bacterial colonies for both groups was 10-20 CFU ml-1. However, there was a significant difference between VAS scores for the two systems (p = 0.000). The mean VAS scores (range 0-10) for patient discomfort for the hygienic sheath group and the plastic bag group were 3.03 and 5.33, respectively. CONCLUSIONS: Based on the design of this study, the alternative barrier provided similar results to those commercially available. Regarding the type of barrier envelope, the hygienic sheath induced less discomfort than the plastic bag.

Accuracy of an equation for estimating age from mandibular third molar development in a Thai population.

PURPOSE: This study assessed the accuracy of age estimates produced by a regression equation derived from lower third molar development in a Thai population. MATERIALS AND METHODS: The first part of this study relied on measurements taken from panoramic radiographs of 614 Thai patients aged from 9 to 20. The stage of lower left and right third molar development was observed in each radiograph and a modified Gat score was assigned. Linear regression on this data produced the following equation: Y=9.309+1.673 mG+0.303S (Y=age; mG=modified Gat score; S=sex). In the second part of this study, the predictive accuracy of this equation was evaluated using data from a second set of panoramic radiographs (539 Thai subjects, 9 to 24 years old). Each subject's age was estimated using the above equation and compared against age calculated from a provided date of birth. Estimated and known age data were analyzed using the Pearson correlation coefficient and descriptive statistics. RESULTS: Ages estimated from lower left and lower right third molar development stage were significantly correlated with the known ages (r=0.818, 0.808, respectively, P≤0.01). 50% of age estimates in the second part of the study fell within a range of error of ±1 year, while 75% fell within a range of error of ±2 years. The study found that the equation tends to estimate age accurately when individuals are 9 to 20 years of age. CONCLUSION: The equation can be used for age estimation for Thai populations when the individuals are 9 to 20 years of age.

Diagnostic performance of cone-beam computed tomography on detection of mechanically-created artificial secondary caries.

PURPOSE: The aim of this study was to compare the diagnostic accuracy of cone-beam computed tomography (CBCT) images and bitewing images in detection of secondary caries. MATERIALS AND METHODS: One hundred and twenty proximal slots of Class II cavities were randomly prepared on human premolar and molar teeth, and restored with amalgam (n=60) and composite resin (n=60). Then, artificial secondary caries lesions were randomly created using round steel No. 4 bur. The teeth were radiographed with a conventional bitewing technique and two CBCT systems; Pax-500ECT and Promax 3D. All images were evaluated by five observers. The area under the receiver operating characteristic (ROC) curve (A(z)) was used to evaluate the diagnostic accuracy. Significant difference was tested using the Friedman test (p value<0.05). RESULTS: The mean A(z) values for bitewing, Pax-500ECT, and Promax 3D imaging systems were 0.882, 0.995, and 0.978, respectively. Significant differences were found between the two CBCT systems and film (p=0.007). For CBCT systems, the axial plane showed the greatest A(z) value. CONCLUSION: Based on the design of this study, CBCT images were better than bitewing radiographs in detection of secondary caries.

Increase of mast cells and tumor angiogenesis in oral squamous cell carcinoma.

BACKGROUND: Although mast cells (MCs) have been implicated in promoting angiogenesis in some malignant tumors, especially of the aerodigestive tract, little is known in oral squamous cell carcinoma (SCC). METHODS: A retrospective study was conducted to elaborate upon the correlation between MCs and tumor angiogenesis in 26 cases of oral SCC, six cases of oral pre-malignant dysplasia, 10 cases of oral hyperkeratosis, and six cases of normal oral mucosa by means of immunohistochemical technique. RESULTS: The MCs in all lesions and normal oral mucosa strongly expressed tryptase. The densities of MCs and microvessels appeared to increase with disease progression. The MC and microvascular counts were significantly higher in oral SCC than in hyperkeratosis and normal oral mucosa (P < 0.05). A significant correlation between MC and microvascular densities was observed in oral SCC (r = 0.5; P = 0.012). CONCLUSIONS: These findings suggest that MCs may upregulate tumor angiogenesis in oral SCC, perhaps via MC tryptase.