Cephalometric Landmark Identification Research Articles

Accuracy of web-based automated versus digital manual cephalometric landmark identification.

The purpose of this study was to assess the accuracy of two web-based automated cephalometric landmark identification and analysis programs. Manual landmark identification using Dolphin Imaging software was used as reference. 105 cephalograms were selected and divided into three groups of 35 subjects each, Class I, II and III. Radiographs were traced using Dolphin imaging software. WebCeph™ (South Korea) and Cephio™ (Poland) were used for the automated cephalometric analysis. Bland-Altman limits of agreement and the concordance correlation coefficient (CCC) were calculated. Kruskal Wallis test was used to compare the accuracy of WebCeph™ and Cephio™ measurements between the three groups. Mann-Whitney U test was used to compare the absolute difference between cephalometric measurements obtained using WebCeph™ and Cephio™. The mean difference (MD) between AI and manually-derived measurements was less than 1mm/degree and ranged from 0.01 to 0.8 except for upper lip protrusion (MD 1.35°), nasolabial angle (MD 5.01°), SN-GoGn (MD 1.41°), Ramus height (MD 1.46°), and IMPA (MD 1.94°). The mean CCC was 0.91 (range 0.60 to 0.96). No statistically significant differences were found between the three malocclusion groups for most of the measurements (P > 0.05). For most of the measurements, automated cephalometric measurements were clinically acceptable. Few differences were found between Webceph™ and Cephio™ for most measurements. Measurements including SNA, SN-PP, IMPA as well as soft tissue measurements require extra consideration and manual adjustment of respective landmarks for higher precision and improved efficiency.

A two-stage regression framework for automated cephalometric landmark detection incorporating semantically fused anatomical features and multi-head refinement loss

Accurate identification and precise localization of cephalometric landmarks provide clinicians with essential insights into craniofacial deformities, aiding in the assessment of treatment strategies for improved patient outcomes. The current methodologies heavily depend on the utilization of multiple CNNs for predicting landmark coordinates, which makes them computationally burdensome and unsuitable for translation to clinical applications. To overcome this limitation, we propose a novel, end-to-end trainable, two-stage regression framework for cephalometric landmark detection. In the initial stage, a single neural network is employed to estimate the locations of all landmarks simultaneously, enabling the identification of potential landmark regions. In the second stage, a semantic fusion block leverages the in-network multi-resolution feature hierarchy to produce high-level semantically rich features. These feature maps are then cropped based on coarsely detected landmark locations and concurrent refinement loss is used to fine-tune and refine the landmark locations. The proposed framework demonstrates the potential for enhancing clinical workflow and treatment outcomes in orthodontics. This is achieved through the utilization of a single CNN backbone augmented with multi-resolution semantically fused anatomical features, which effectively enhances representation learning in a computationally efficient manner. The performance of the proposed framework is evaluated on two publicly available anatomical landmark datasets. The experimental results demonstrate that our framework achieves a state-of-the-art detection accuracy of 87.17% within the clinically accepted range of 2 mm. The source code and the pre-trained weights are made publicly available at this https://github.com/manwaarkhd/CEPHMark-Net, promoting reproducibility and enabling further advancements.

Accuracy of and dental students' preferences toward manual and digital cephalometric landmark identification: A randomized cross-over study.

To evaluate dental students' perceptions of manual and digital cephalometric landmark identification methods based on their preferences, difficulty level, and procedure time required to provide insights into the future of dental education, considering incorporating digital technology in dental schools. Fifty-five second-year dental students were randomly divided into two groups: (1) group A, students who performed manual landmark identification first, followed by digital method; and (2) group B, students who performed digital method first, followed by manual method. The duration of the procedure was recorded. Subsequently, all students completed a questionnaire regarding the difficulty they experienced using a visual analog scale and their preferences. Landmark identification accuracy was measured. Digital landmark identification was preferred by 93% of students. The mean procedure time for digital method was significantly lower than that of manual method (13.00±5.60vs. 9.70±4.60; p=0.002). Group B completed manual and digital methods in a shorter time than group A. Group A experienced less difficulty with manual procedure than group B. However, statistically significant differences were not observed in the difficulty level of digital technique. A statistically significant difference in the mean accuracy was shown in favor of the manual method. However, this difference is clinically insignificant (p=0.001). Students considered digital method to be effective for learning and preferred it over manual method. Furthermore, digital landmark identification demonstrated better performance and was faster than manual method, suggesting that this must be incorporated in undergraduate dental education.

Assessment of landmark detection in cephalometric radiographs with different conditions of brightness and contrast using the an artificial intelligence software.

To evaluate the reliability and reproducibility of an artificial intelligence (AI) software in identifying cephalometric points on lateral cephalometric radiographs considering four settings of brightness and contrast. Brightness and contrast of 30 lateral cephalometric radiographs were adjusted into four different settings. Then, the control examiner (ECont), the calibrated examiner (ECal), and the CEFBOT AI software (AIs) each marked 19 cephalometric points on all radiographs. Reliability was assessed with a second analysis of the radiographs 15 days after the first one. Statistical significance was set at p < 0.05. Reliability of landmark identification was excellent for the human examiners and the AIs regardless of the type of brightness and contrast setting (mean intraclass correlation coefficient >0.89). When ECont and ECal were compared for reproducibility, there were more cephalometric points with significant differences on the x-axis of the image with the highest contrast and the lowest brightness, namely N(p = 0.033), S(p = 0.030), Po(p < 0.001), and Pog'(p = 0.012). Between ECont and AIs, there were more cephalometric points with significant differences on the image with the highest contrast and the lowest brightness, namely N(p = 0.034), Or(p = 0.048), Po(p < 0.001), A(p = 0.042), Pog'(p = 0.004), Ll(p = 0.005), Ul(p < 0.001), and Sn(p = 0.001). While the reliability of the AIs for cephalometric landmark identification was rated as excellent, low brightness and high contrast seemed to affect its reproducibility. The experienced human examiner, on the other hand, did not show such faulty reproducibility; therefore, the AIs used in this study is an excellent auxiliary tool for cephalometric analysis, but still depends on human supervision to be clinically reliable.

Automatic identification of posteroanterior cephalometric landmarks using a novel deep learning algorithm: a comparative study with human experts

This study aimed to propose a fully automatic posteroanterior (PA) cephalometric landmark identification model using deep learning algorithms and compare its accuracy and reliability with those of expert human examiners. In total, 1032 PA cephalometric images were used for model training and validation. Two human expert examiners independently and manually identified 19 landmarks on 82 test set images. Similarly, the constructed artificial intelligence (AI) algorithm automatically identified the landmarks on the images. The mean radial error (MRE) and successful detection rate (SDR) were calculated to evaluate the performance of the model. The performance of the model was comparable with that of the examiners. The MRE of the model was 1.87 ± 1.53 mm, and the SDR was 34.7%, 67.5%, and 91.5% within error ranges of < 1.0, < 2.0, and < 4.0 mm, respectively. The sphenoid points and mastoid processes had the lowest MRE and highest SDR in auto-identification; the condyle points had the highest MRE and lowest SDR. Comparable with human examiners, the fully automatic PA cephalometric landmark identification model showed promising accuracy and reliability and can help clinicians perform cephalometric analysis more efficiently while saving time and effort. Future advancements in AI could further improve the model accuracy and efficiency.

Evaluation of AI Model for Cephalometric Landmark Classification (TG Dental).

The accuracy of cephalometric landmark identification for malocclusion classification is essential for diagnosis and treatment planning. Identifying these landmarks is often complex and time-consuming for orthodontists. An AI model for classification was recently developed. This model was investigated based on current regulatory considerations as a result of the strict regulations on software systems and the lack of information on artificial intelligence (AI) requirements in this publication. The platform developed by the ITU/WHO for AI is used to assess the models of the application. The auditing procedure assessed the development process concerning medical device regulations, data protection regulations, and ethical considerations. Upon that, the major tasks during the development were evaluated, such as qualification, annotation procedure, and data set attributes. The AI models were investigated under consideration of technical, clinical, regulatory, and ethical considerations. The risk to the patient and user's health can be considered low according to the International Medical Device Regulators Forum (IMDRF) definition. This application facilitates the decision and planning of malocclusion treatment based on lateral cephalograms without cephalometric landmarks. It is comparable with common standards in orthodontic diagnosis.

Application of Artificial Intelligence (AI) in a Cephalometric Analysis: A Narrative Review.

In recent years, the application of artificial intelligence (AI) has become more and more widespread in medicine and dentistry. It may contribute to improved quality of health care as diagnostic methods are getting more accurate and diagnostic errors are rarer in daily medical practice. The aim of this paper was to present data from the literature on the effectiveness of AI in orthodontic diagnostics based on the analysis of lateral cephalometric radiographs. A review of the literature from 2009 to 2023 has been performed using PubMed, Medline, Scopus and Dentistry & Oral Sciences Source databases. The accuracy of determining cephalometric landmarks using widely available commercial AI-based software and advanced AI algorithms was presented and discussed. Most AI algorithms used for the automated positioning of landmarks on cephalometric radiographs had relatively high accuracy. At the same time, the effectiveness of using AI in cephalometry varies depending on the algorithm or the application type, which has to be accounted for during the interpretation of the results. In conclusion, artificial intelligence is a promising tool that facilitates the identification of cephalometric landmarks in everyday clinical practice, may support orthodontic treatment planning for less experienced clinicians and shorten radiological examination in orthodontics. In the future, AI algorithms used for the automated localisation of cephalometric landmarks may be more accurate than manual analysis.

Human examination and artificial intelligence in cephalometric landmark detection-is AI ready to take over?

To compare the precision of two cephalometric landmark identification methods, namely a computer-assisted human examination software and an artificial intelligence program, based on South African data. This retrospective quantitative cross-sectional analytical study utilized a data set consisting of 409 cephalograms obtained from a South African population. 19 landmarks were identified in each of the 409 cephalograms by the primary researcher using the two programs [(409 cephalograms x 19 landmarks) x 2 methods = 15,542 landmarks)]. Each landmark generated two coordinate values (x, y), making a total of 31,084 landmarks. Euclidean distances between corresponding pairs of observations was calculated. Precision was determined by using the standard deviation and standard error of the mean. The primary researcher acted as the gold-standard and was calibrated prior to data collection. The inter- and intrareliability tests yielded acceptable results. Variations were present in several landmarks between the two approaches; however, they were statistically insignificant. The computer-assisted examination software was very sensitive to several variables. Several incidental findings were also discovered. Attempts were made to draw valid comparisons and conclusions. There was no significant difference between the two programs regarding the precision of landmark detection. The present study provides a basis to: (1) support the use of automatic landmark detection to be within the range of computer-assisted examination software and (2) determine the learning data required to develop AI systems within an African context.

Artificial Intelligence and Machine Learning for Automated Cephalometric Landmark Identification: A Meta-Analysis Previewed by a Systematic Review.

Digital dentistry has become an integral part of our practice today, with artificial intelligence (AI) playing the predominant role. The present systematic review was intended to detect the accuracy of landmarks identified cephalometrically using machine learning and artificial intelligence and compare the same with the manual tracing (MT) group. According to the PRISMA-DTA guidelines, a scoping evaluation of the articles was performed. Electronic databases like Doaj, PubMed, Scopus, Google Scholar, and Embase from January 2001 to November 2022 were searched. Inclusion and exclusion criteria were applied, and 13 articles were studied in detail. Six full-text articles were further excluded (three articles did not provide a comparison between manual tracing and AI for cephalometric landmark detection, and three full-text articles were systematic reviews and meta-analyses). Finally, seven articles were found appropriate to be included in this review. The outcome of this systematic review has led to the conclusion that AI, when employed for cephalometric landmark detection, has shown extremely positive and promising results as compared to manual tracing.

Artificial Intelligence in Pediatric Dentistry – A Systematic Review

Abstract Background: Artificial intelligence (AI) refers to the development of computer systems capable of performing tasks that normally require human intelligence. AI and its subsets, machine learning and deep learning, have been incorporated into several aspects of dentistry including pediatric dentistry. However, there is a lack of documentation and analysis of the current applications of AI in pediatric dentistry. Aim: The aim of this systematic review was to evaluate the effectiveness of AI as a diagnostic tool in pediatric dentistry. Materials and Methods: The literature for this paper was identified by performing a thorough search in electronic databases such as PubMed, Google Scholar, and Cochrane Library from the years 2011 to 2021. The following keywords and Boolean operators were used: AI AND pediatric dentistry, artificial neural networks AND pediatric dentistry, convolutional neural networks AND pediatric dentistry, and machine learning AND pediatric dentistry. After applying appropriate inclusion and exclusion criteria, 13 articles were selected, fully read, and systematically analyzed as per a specific research question. Results: Among the 13 selected articles, it was found that AI is a useful tool for dental diagnosis/classification, cephalometric landmark identification, identification of early childhood caries patterns, chronological age assessment in children, assessment of facial attractiveness in cleft patients, dental plaque detection, and oral health education. Conclusion: The selected articles indicate that AI is an effective diagnostic tool and has the potential for assisting several aspects of pediatric dentistry. However, further studies are required to assess the clinical effectiveness of these AI models.

Verification of Convolutional Neural Network Cephalometric Landmark Identification

Introduction: The mass-harvesting of digitized medical data has prompted their use as a clinical and research tool. The purpose of this study was to compare the accuracy and reliability of artificial intelligence derived cephalometric landmark identification with that of human observers. Methods: Ten pre-treatment digital lateral cephalometric radiographs were randomly selected from a university post-graduate clinic. The x- and y-coordinates of 21 (i.e., 42 points) hard and soft tissue landmarks were identified by 6 specialists, 19 residents, 4 imaging technicians, and a commercially available convolutional neural network artificial intelligence platform (CephX, Orca Dental, Hertzylia, Israel). Wilcoxon, Spearman and Bartlett tests were performed to compare agreement of human and AI landmark identification. Results: Six x- or y-coordinates (14.28%) were found to be statistically different, with only one being outside the 2 mm range of acceptable error, and with 97.6% of coordinates found to be within this range. Conclusions: The use of convolutional neural network artificial intelligence as a tool for cephalometric landmark identification was found to be highly accurate and can serve as an aid in orthodontic diagnosis.

Development, Application, and Performance of Artificial Intelligence in Cephalometric Landmark Identification and Diagnosis: A Systematic Review.

This study aimed to analyze the existing literature on how artificial intelligence is being used to support the identification of cephalometric landmarks. The systematic analysis of literature was carried out by performing an extensive search in PubMed/MEDLINE, Google Scholar, Cochrane, Scopus, and Science Direct databases. Articles published in the last ten years were selected after applying the inclusion and exclusion criteria. A total of 17 full-text articles were systematically appraised. The Cochrane Handbook for Systematic Reviews of Interventions (CHSRI) and Newcastle-Ottawa quality assessment scale (NOS) were adopted for quality analysis of the included studies. The artificial intelligence systems were mainly based on deep learning-based convolutional neural networks (CNNs) in the included studies. The majority of the studies proposed that AI-based automatic cephalometric analyses provide clinically acceptable diagnostic performance. They have worked remarkably well, with accuracy and precision similar to the trained orthodontist. Moreover, they can simplify cephalometric analysis and provide a quick outcome in practice. Therefore, they are of great benefit to orthodontists, as with these systems they can perform tasks more efficiently.

Trends and Application of Artificial Intelligence Technology in Orthodontic Diagnosis and Treatment Planning—A Review

Artificial intelligence (AI) is a new breakthrough in technological advancements based on the concept of simulating human intelligence. These emerging technologies highly influence the diagnostic process in the field of medical sciences, with enhanced accuracy in diagnosis. This review article intends to report on the trends and application of AI models designed for diagnosis and treatment planning in orthodontics. A data search for the original research articles that were published over the last 22 years (from 1 January 2000 until 31 August 2022) was carried out in the most renowned electronic databases, which mainly included PubMed, Google Scholar, Web of Science, Scopus, and Saudi Digital Library. A total of 56 articles that met the eligibility criteria were included. The research trend shows a rapid increase in articles over the last two years. In total: 17 articles have reported on AI models designed for the automated identification of cephalometric landmarks; 12 articles on the estimation of bone age and maturity using cervical vertebra and hand-wrist radiographs; two articles on palatal shape analysis; seven articles for determining the need for orthodontic tooth extractions; two articles for automated skeletal classification; and 16 articles for the diagnosis and planning of orthognathic surgeries. AI is a significant development that has been successfully implemented in a wide range of image-based applications. These applications can facilitate clinicians in diagnosing, treatment planning, and decision-making. AI applications are beneficial as they are reliable, with enhanced speed, and have the potential to automatically complete the task with an efficiency equivalent to experienced clinicians. These models can prove as an excellent guide for less experienced orthodontists.

A semi-supervised learning approach for automated 3D cephalometric landmark identification using computed tomography.

Identification of 3D cephalometric landmarks that serve as proxy to the shape of human skull is the fundamental step in cephalometric analysis. Since manual landmarking from 3D computed tomography (CT) images is a cumbersome task even for the trained experts, automatic 3D landmark detection system is in a great need. Recently, automatic landmarking of 2D cephalograms using deep learning (DL) has achieved great success, but 3D landmarking for more than 80 landmarks has not yet reached a satisfactory level, because of the factors hindering machine learning such as the high dimensionality of the input data and limited amount of training data due to the ethical restrictions on the use of medical data. This paper presents a semi-supervised DL method for 3D landmarking that takes advantage of anonymized landmark dataset with paired CT data being removed. The proposed method first detects a small number of easy-to-find reference landmarks, then uses them to provide a rough estimation of the all landmarks by utilizing the low dimensional representation learned by variational autoencoder (VAE). The anonymized landmark dataset is used for training the VAE. Finally, coarse-to-fine detection is applied to the small bounding box provided by rough estimation, using separate strategies suitable for the mandible and the cranium. For mandibular landmarks, patch-based 3D CNN is applied to the segmented image of the mandible (separated from the maxilla), in order to capture 3D morphological features of mandible associated with the landmarks. We detect 6 landmarks around the condyle all at once rather than one by one, because they are closely related to each other. For cranial landmarks, we again use the VAE-based latent representation for more accurate annotation. In our experiment, the proposed method achieved a mean detection error of 2.88 mm for 90 landmarks using only 15 paired training data.

Comparison of AudaxCeph®’s fully automated cephalometric tracing technology to a semi-automated approach by human examiners

To compare the reliability of cephalometric landmark identification by an automated tracing software based on convolutional neural networks to human tracers. Sixty cephalograms were traced by two board-certified orthodontists and AudaxCeph®'s artificial intelligence software. The following thirteen landmarks were identified in each tracing: Sella, Nasion, A point, B point, Porion, Menton, Pogonion, Orbitale, Gonion, Upper Central Incisor Incisal Edge (U1 Tip), Upper Central Incisor Root Apex (U1 apex), Lower Central Incisor Incisal Edge (L1 Tip), Lower Central Incisor Root Apex (L1 apex). An x-y axis was positioned in the bottom left corner of each cephalogram, and the x- and y-coordinates for the landmarks were exported into Excel. Distributions of landmarks (X, Y, radial distance) were compared using t-tests of equivalence with a 2mm equivalence bound. These compared the AI position to the two orthodontists-and the orthodontists' reliability by comparing equivalence against each other. There was no statistical difference between the orthodontists and AudaxCeph®'s automatic tracing software except for the x- and y-dimension of Porion and the y-dimension of L1 apex. The two orthodontists had good intra-examiner reliability with no statistical difference found when comparing them. AudaxCeph®'s automated cephalometric tracing software is a good adjunctive tool to use when diagnosing and treatment planning orthodontic cases.

Cephalometric Analysis in Orthodontics Using Artificial Intelligence—A Comprehensive Review

Artificial intelligence (AI) is a branch of science concerned with developing programs and computers that can gather data, reason about it, and then translate it into intelligent actions. AI is a broad area that includes reasoning, typical linguistic dispensation, machine learning, and planning. In the area of medicine and dentistry, machine learning is currently the most widely used AI application. This narrative review is aimed at giving an outline of cephalometric analysis in orthodontics using AI. Latest algorithms are developing rapidly, and computational resources are increasing, resulting in increased efficiency, accuracy, and reliability. Current techniques for completely automatic identification of cephalometric landmarks have considerably improved efficiency and growth prospects for their regular use. The primary considerations for effective orthodontic treatment are an accurate diagnosis, exceptional treatment planning, and good prognosis estimation. The main objective of the AI technique is to make dentists' work more precise and accurate. AI is increasingly being used in the area of orthodontic treatment. It has been evidenced to be a time-saving and reliable tool in many ways. AI is a promising tool for facilitating cephalometric tracing in routine clinical practice and analyzing large databases for research purposes.

Effectiveness of Human-Artificial Intelligence Collaboration in Cephalometric Landmark Detection.

Detection of cephalometric landmarks has contributed to the analysis of malocclusion during orthodontic diagnosis. Many recent studies involving deep learning have focused on head-to-head comparisons of accuracy in landmark identification between artificial intelligence (AI) and humans. However, a human–AI collaboration for the identification of cephalometric landmarks has not been evaluated. We selected 1193 cephalograms and used them to train the deep anatomical context feature learning (DACFL) model. The number of target landmarks was 41. To evaluate the effect of human–AI collaboration on landmark detection, 10 images were extracted randomly from 100 test images. The experiment included 20 dental students as beginners in landmark localization. The outcomes were determined by measuring the mean radial error (MRE), successful detection rate (SDR), and successful classification rate (SCR). On the dataset, the DACFL model exhibited an average MRE of 1.87 ± 2.04 mm and an average SDR of 73.17% within a 2 mm threshold. Compared with the beginner group, beginner–AI collaboration improved the SDR by 5.33% within a 2 mm threshold and also improved the SCR by 8.38%. Thus, the beginner–AI collaboration was effective in the detection of cephalometric landmarks. Further studies should be performed to demonstrate the benefits of an orthodontist–AI collaboration.

Effects of facemask therapy in the treatment of skeletal class III malocclusion in Vietnamese children

Sparse data are available on the effects of facemask therapy in treatment of skeletal class III malocclusion in Vietnamese children. The purpose of this study was to investigate the skeletal, dental, and soft-tissue profile changes following facemask treatment in children with skeletal class III malocclusion. Twenty-four children (7-12 years-old) with skeletal class III malocclusion treated with facemask appliance were included in this study. All lateral cephalograms taken before and after treatment had been manually traced, followed by cephalometric landmark identification and parameter measurements. Significant skeletal, dental and soft tissue profile changes after treatment were recorded and statistically evaluated by paired t test at a 0.05 significance level. After treatment, the maxilla moved forward (SNA increased 1.53 degrees and A–Y increased 1.93 mm; p&lt;0.001). The mandible rotated backwards and downwards (SNB decreased 1.35 degrees; B–Y decreased 1.12mm; p&lt;0.001). These movements in the maxilla and mandible caused a significant improvement in intermaxillary sagittal relationship (ANB increased 2.89degrees; the convexity angle increased 4.7 degrees; Wits appraisal increased 3.77mm; p&lt;0.001). The maxillary incisors moved forward (4.57 degrees). The improvement in overjet was 5.83 mm. The change in Prn-Y and Ls–E measurement was 2.46mm and 2.72mm, respectively. Our results showed that facemask was highly effective for treating skeletal class III malocclusion and improving facial esthetics. Facemask treatment is a highly effective for skeletal class III malocclusion because it leads to positive changes in the jaws, teeth and soft tissue after 1-year of treatment.

Assessment of automatic cephalometric landmark identification using artificial intelligence.

To compare the accuracy of cephalometric landmark identification between artificial intelligence (AI) deep learning convolutional neural networks (CNN) You Only Look Once, Version 3 (YOLOv3) algorithm and the manually traced (MT) group. The American Association of Orthodontists Federation (AAOF) Legacy Denver collection was used to obtain 110 cephalometric images for this study. Lateral cephalograms were digitized and traced by a calibrated senior orthodontic resident using Dolphin Imaging. The same images were uploaded to AI software Ceppro DDH Inc The Cartesian system of coordinates with Sella as the reference landmark was used to extract x- and y-coordinates for 16 cephalometric points: Nasion (Na), A point, B point, Menton (Me), Gonion (Go), Upper incisor tip, Lower incisor tip, Upper incisor apex, Lower incisor apex, Anterior Nasal Spine (ANS), Posterior Nasal Spine (PNS), Pogonion (Pg), Pterigomaxillary fissure point (Pt), Basion (Ba), Articulare (Art) and Orbitale (Or). The mean distances were assessed relative to the reference value of 2mm. Student paired t-tests at significance level of P<.05 were used to compare the mean differences in each of the x- and y-components. SPSS (IBM-vs. 27.0) software was used for the data analysis. There was no statistical difference for 12 out of 16 points when analysing absolute differences between MT and AI groups. AI may increase efficiency without compromising accuracy with cephalometric tracings in routine clinical practice and in research settings.

Fully automated identification of cephalometric landmarks for upper airway assessment using cascaded convolutional neural networks.

The aim of the study was to evaluate the accuracy of a cascaded two-stage convolutional neural network (CNN) model in detecting upper airway (UA) soft tissue landmarks in comparison with the skeletal landmarks on the lateral cephalometric images. The dataset contained 600 lateral cephalograms of adult orthodontic patients, and the ground-truth positions of 16 landmarks (7 skeletal and 9 UA landmarks) were obtained from 500 learning dataset. We trained a UNet with EfficientNetB0 model through the region of interest-centred circular segmentation labelling process. Mean distance errors (MDEs, mm) of the CNN algorithm was compared with those from human examiners. Successful detection rates (SDRs, per cent) assessed within 1-4 mm precision ranges were compared between skeletal and UA landmarks. The proposed model achieved MDEs of 0.80 ± 0.55 mm for skeletal landmarks and 1.78 ± 1.21 mm for UA landmarks. The mean SDRs for UA landmarks were 72.22 per cent for 2 mm range, and 92.78 per cent for 4 mm range, contrasted with those for skeletal landmarks amounting to 93.43 and 98.71 per cent, respectively. As compared with mean interexaminer difference, however, this model showed higher detection accuracies for geometrically constructed UA landmarks on the nasopharynx (AD2 and Ss), while lower accuracies for anatomically located UA landmarks on the tongue (Td) and soft palate (Sb and St). The proposed CNN model suggests the availability of an automated cephalometric UA assessment to be integrated with dentoskeletal and facial analysis.