Development of an artificial intelligence-based algorithm to classify images acquired with an intraoral scanner of individual molar teeth into three categories

Diagnostic validity of ICDAS clinical criteria on digital 3D models

OBJECTIVE

To assess the diagnostic validity of ICDAS clinical criteria on 3D dental models from intraoral scanning.

METHODS

This is a retrospective analysis on part of the baseline sample collected in a cohort study and included 73 participants (12-19 years) from Denmark and Greece. The assessment was made by visual inspection, and then by visual inspection associated with radiographs. All participants were scanned with TRIOS 4 which uses white light to obtain the 3D models with tooth color, as well as blue light source (415 nm) for fluorescence. The 3D models' evaluation was conducted using tooth-color texture and subsequently fluorescence. Two scores were obtained for the 3D model examination: i) ICDAS based on tooth-color information; ii) ICDAS based on tooth-color information supplemented with fluorescence. For the analysis, weighted kappa, sensitivity (SE), specificity (SP) and accuracy (ACC) were calculated.

RESULTS

Regarding all lesions the values for SE, SP, and ACC were respectively 0.804, 0.801, and 0.802 for tooth-color, and 0.819, 0.808, and 0.810 for tooth-color supplemented with fluorescence. In terms of accuracy parameters for moderate-extensive lesions, the values for SE, SP, and ACC for tooth color were 0.709, 0.948, and 0.944, while for fluorescence, they were 0.815, 0.937, and 0.934.

CONCLUSION

Caries assessment with ICDAS criteria on 3D dental models produces reliable scores. Visual caries analysis using 3D scans demonstrates commendable diagnostic accuracy and reasonable consistency with traditional methods. The use of intraoral scanners may be beneficial in evaluating occlusal caries.

CLINICAL SIGNIFICANCE

The importance of this study is to prove the diagnostic accuracy of caries lesions diagnosis made using and intraoral scanner and to offer greater confidence to professionals who use this diagnosis tool in their daily clinical practice. Intraoral scanners demonstrate to be an accurate tool for diagnosing occlusal caries.

The Potential of Artificial Intelligence in Prosthodontics: A Comprehensive Review

Prosthodontics is a dental subspecialty that includes the preparation of dental prosthetics for missing or damaged teeth. It increasingly uses computer-assisted technologies for planning and preparing dental prosthetics. This study aims to present the findings from a systematic review of publications on artificial intelligence (AI) in prosthodontics to identify current trends and future opportunities. The review question was "What are the applications of AI in prosthodontics and how good is their performance in prosthodontics?" Electronic searching in the Web of Science, ScienceDirect, PubMed, and Cochrane Library was conducted. The search was limited to full text from January 2012 to January 2024. Quadas-2 was used for assessing quality and potential risk of bias for the selected studies. A total of 1925 studies were identified in the initial search. After removing the duplicates and applying exclusion criteria, a total of 30 studies were selected for this review. Results of the Quadas-2 assessment of included studies found that a total of 18.3% of studies were identified as low risk of bias studies, whereas 52.6% and 28.9% of included studies were identified as studies with high and unclear risk of bias, respectively. Although they are still developing, AI models have already shown promise in the areas of dental charting, tooth shade selection, automated restoration design, mapping the preparation finishing line, manufacturing casting optimization, predicting facial changes in patients wearing removable prostheses, and designing removable partial dentures.

Automatic removal of soft tissue from 3D dental photo scans; an important step in automating future forensic odontology identification

The potential of intraoral 3D photo scans in forensic odontology identification remains largely unexplored, even though the high degree of detail could allow automated comparison of ante mortem and post mortem dentitions. Differences in soft tissue conditions between ante- and post mortem intraoral 3D photo scans may cause ambiguous variation, burdening the potential automation of the matching process and underlining the need for limiting inclusion of soft tissue in dental comparison. The soft tissue removal must be able to handle dental arches with missing teeth, and intraoral 3D photo scans not originating from plaster models. To address these challenges, we have developed the grid-cutting method. The method is customisable, allowing fine-grained analysis using a small grid size and adaptation of how much of the soft tissues are excluded from the cropped dental scan. When tested on 66 dental scans, the grid-cutting method was able to limit the amount of soft tissue without removing any teeth in 63/66 dental scans. The remaining 3 dental scans had partly erupted third molars (wisdom teeth) which were removed by the grid-cutting method. Overall, the grid-cutting method represents an important step towards automating the matching process in forensic odontology identification using intraoral 3D photo scans.

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.

3D superimposition of human dentition contours in personal identification: A preliminary study

The human permanent dentition has been commonly used for personal identification due to its uniqueness. Limited research, however, is conducted using 3D digital dental models. We propose to develop a new 3D superimposition method using the contours of human dentition and to further evaluate its feasibility. A total of 270 intraoral scan models were collected from 135 subjects. After a one-year interval, 52 subjects were chosen at random and the secondary intraoral scan models were obtained. The dentition contours of the first and secondary models were extracted to form a resource dataset and a test dataset. Through the application of the iterative nearest point (ICP) algorithm, the test dataset was registered with the resource dataset, and the root mean square error (RMSE) values of the point-to-point distances were calculated. 104 genuine pairs and 13,936 imposter pairs were generated, and in this study, the registration accuracy was 100%. The difference between mean RMSE values for the genuine pair (0.20 ± 0.06 mm) and the minimum RMSE value for the imposter pair (0.83 ± 0.06 mm) was significant in the maxillary arch (p < 0.05). Similarly, in the mandibular arch, the difference between mean RMSE values for the genuine pair (0.22 ± 0.07 mm) and the minimum RMSE value for the imposter pair (0.85 ± 0.08 mm) was significant (p < 0.05). The difference between the RMSE value for the genuine pair in the maxillary and the mandibular arch was significant (p < 0.05). This study indicated the feasibility of dentition contour-based model superimposition and could be considered for personal identification in the future.

Assessment of fully digitalized workflow for implant-prosthetic rehabilitation in temporomandibular disorders patients: A clinical study

Digitalized workflow eliminates the need for the tray, impression materials, its decontamination, packaging and shipping, pouring with plaster, cast fabrication, mounting in an articulator, reducing storage spaces, and the risks of any loss or fracture of the plaster model is overcome by archiving on the computer. This clinical investigation aimed to evaluate the effectiveness of the fully digitalized rehabilitation [implant-supported prosthesis] method in partially edentulous patients and with TMD, using advanced software. Twelve patients requiring implant-supported prosthesis in the mandibular molar area with Temporomandibular disorders [TMD] were selected. The fully digitalized rehabilitation method with advanced software was used for rehabilitation. For each subject, Optical impressions, CBCT scan, and Digital recording of jaw movement data. Guided implant surgery and digitalized prosthetic rehabilitation; were performed. The effectiveness of the digitalized workflow was assessed by evaluating the changes in the joint symptoms before and after the end of the treatment, changes in the electromyographic tracings, the precision of the prosthetic artefact, assessed through the amount of chair adjustment operating time and the number of retouching/ modifications to be carried out before the completion of the work. The results showed that the mean operative time required in 12 patients was 9.42 min, significantly less than the time recorded in previous studies when the medium mean was 16.00 min. The mean number of touch-ups [adjustments] was less than 3, most of which were on the interproximal surfaces. There were no significant changes recorded in the electromyography tracings. There were also no changes in joint symptoms. It was found that this way of working was entirely reliable and significantly reduced operating times and the number of appointments. Digital flow is beneficial ei dysfunctional patients, not about improvements in temporomandibular symptoms but in times of operability and prosthetic retouching.

Artificial intelligence in forensic medicine and forensic dentistry

This review article aims to highlight the current possibilities for applying Artificial Intelligence in modern forensic medicine and forensic dentistry and present the advantages and disadvantages of its use. For this purpose, the relevant academic literature was searched using PubMed, Web of Science and Scopus. The application of Artificial Intelligence in forensic medicine and forensic dentistry is still in its early stages. However, the possibilities are great, and the future will show what is applicable in daily practice. Artificial Intelligence will improve the accuracy and efficiency of work in forensic medicine and forensic dentistry; it can automate some tasks; and enhance the quality of evidence. Disadvantages of the application of Artificial Intelligence may be related to discrimination, transparency, accountability, privacy, security, ethics and others. Artificial Intelligence systems should be used as a support tool, not as a replacement for forensic experts.

A modified scan technique for multiple abutment teeth using the trim and lock function

OBJECTIVES

To describe a new protocol for digital scanning of multiple abutment teeth using the trim and lock software tools.

METHODS

A reverse workflow technique was used. Scanning was performed with the interim restoration in position. The abutment teeth were then trimmed from the scan. The retraction cord or interim restoration from either the first mesial or distal abutment tooth was removed and only that tooth was scanned, allowing the dentist to easily manage gingival displacement and keep the tooth dry from crevicular fluid and saliva. Consequently, the preparation margin remained visible and uncontaminated during the scan. The adjacent abutment teeth detected in the scan were deleted from it, and the scan was then locked using a tool of the scanning software. Next, the retraction cord or interim restoration of the next abutment tooth was removed, and only that tooth was scanned. The procedure was repeated until all prepared teeth were individually scanned.

RESULTS

The technique presented here facilitated the scanning of multiple abutment teeth in a simple and predictable way by utilizing the trim and lock surface tools of the scanning software and helped in avoiding closure of the gingival crevice.

CONCLUSIONS

Splitting the scan for a complex case with multiple abutment teeth allows reliable 3D acquisition of the finish line of each abutment tooth. Therefore, this technique simplifies the full-arch intraoral scanning process and can improve treatment efficiency.

CLINICAL SIGNIFICANCE

The trim and lock tool allows scanning of each prepared abutment tooth separately, transforming a full-arch impression into multiple single scans. This technique helps to easily manage gingival displacement and maintain an uncontaminated and dry preparation margin during the scan.

Texture-Based Neural Network Model for Biometric Dental Applications

BACKGROUND

The aim is to classify dentition using a novel texture-based automated convolutional neural network (CNN) for forensic and prosthetic applications.

METHODS

Natural human teeth (n = 600) were classified, cleaned, and inspected for exclusion criteria. The teeth were scanned with an intraoral scanner and identified using a texture-based CNN in three steps. First, through preprocessing, teeth images were segmented by extracting the front-facing region of the teeth. Then, texture features were extracted from the segmented teeth images using the discrete wavelet transform (DWT) method. Finally, deep learning-based enhanced CNN models were used to identify these images. Several experiments were conducted using five different CNN models with various batch sizes and epochs, with and without augmented data.

RESULTS

Based on experiments with five different CNN models, the highest accuracy achieved was 0.8 and the precision was 0.8 with a loss value of 0.9, a batch size of 32, and 250 epochs. A comparison of deep learning models with different parameters showed varied accuracy between the different classes of teeth.

CONCLUSION

The accuracy of the point-based CNN method was promising. This texture-identification method will pave the way for many forensic and prosthodontic applications and will potentially help improve the precision of dental biometrics.