• The korean journal of orthodontics
• /
• v.49 no.1
• /
• pp.32-40
• /
• 2019

Objective: Precise identification of landmarks on posteroanterior (PA) cephalograms is necessary when evaluating lateral problems such as facial asymmetry. The aim of the present study was to investigate whether the use of lateral (LA) cephalograms can reduce errors in landmark identification on PA cephalograms. Methods: Five examiners identified 16 landmarks (Cg, N, ANS, GT, Me, RO, Lo, FM, Z, Or, Zyg, Cd, NC, Ms, M, and Ag) on 32 PA cephalograms with and without LA cephalograms at the same time. The positions of the landmarks were recorded and saved in the horizontal and vertical direction. The mean errors and standard deviation of landmarks location according to the use of LA cephalograms were compared for each landmark. Results: Relatively small errors were found for ANS, Me, Ms, and Ag, while relatively large errors were found for N, GT, Z, Or, and Cd. No significant difference was found between the horizontal and vertical errors for Z and Or, while large vertical errors were found for N, GT, and Cd. The value of identification error was lower when the landmarks were identified using LA cephalograms. Statistically significant error reductions were found at N and Cd with LA cephalograms, especially in the vertical direction. Conclusions: The use of LA cephalograms during identification of landmarks on PA cephalograms could help reduce identification errors.

• The korean journal of orthodontics
• /
• v.49 no.1
• /
• pp.41-48
• /
• 2019

Objective: This in-vivo study aimed to compare landmark identification errors in anteroposterior (AP) and posteroanterior (PA) cephalograms generated from cone-beam computed tomography (CBCT) scan data in order to examine the feasibility of using AP cephalograms in clinical settings. Methods: AP and PA cephalograms were generated from CBCT scans obtained from 25 adults. Four experienced and four inexperienced examiners were selected depending on their experience levels in analyzing frontal cephalograms. They identified six cephalometric landmarks on AP and PA cephalograms. The errors incurred in positioning the cephalometric landmarks on the AP and PA cephalograms were calculated by using the straight-line distance and the horizontal and vertical components as parameters. Results: Comparison of the landmark identification errors in CBCT-generated frontal cephalograms revealed that landmark-dependent differences were greater than experienceor projection-dependent differences. Comparisons of landmark identification errors in the horizontal and vertical directions revealed larger errors in identification of the crista galli and anterior nasal spine in the vertical direction and the menton in the horizontal direction, in comparison with the other landmarks. Comparison of landmark identification errors between the AP and PA projections in CBCT-generated images revealed a slightly higher error rate in the AP projections, with no inter-examiner differences. Statistical testing of the differences in landmark identification errors between AP and PA cephalograms showed no statistically significant differences for all landmarks. Conclusions: The reproducibility of CBCT-generated AP cephalograms is comparable to that of PA cephalograms; therefore, AP cephalograms can be generated reliably from CBCT scan data in clinical settings.

• The korean journal of orthodontics
• /
• v.32 no.2 s.91
• /
• pp.79-89
• /
• 2002

The purpose of this study is to evaluate the reproducibility and errors in landmark identification of conventional lateral cephalometric radiography and digital lateral cephalometric radiography. Fifteen conventional lateral cephalometric radiographs and fifteen digital lateral cephalometric radiographs were selected in adults with no considerations on sex and craniofacial forms. Each landmark was identified and expressed as the coordinate (x, y). The landmarks were classified into 3 groups. The landmarks of the first identification was T1, identification after one week was T2, and identification after one month was T3. The mean and standard deviation of identification errors between replicates were calculated according to the x and y coordinates. The errors between first identification and second identification were expressed as T2-T1(x), T2-T1(y) and those between first identification and third identification were expressed as T3-T1(x), T2-T1(y). Each was divided into conventional lateral cephalometric radiography and digital lateral cephalometric radiography. The independent t- test was used for statistical analysis of identification errors for the evaluation of reproducibility. The results of this study were as follows ; 1. Generally, the mean and standard deviation of landmark identification errors in digital lateral cephalometric radiography was smaller than those of conventional lateral cephalometric radiography. 2. Only a few landmarks showed statistically significant difference in identification error between conventional lateral cephalometric radiography and digital lateral cephalometric radiography. 3. The enhancement of image quality didn't guarantee decrease in landmark identification error and didn't affect tendency of landmark identification error.

• The korean journal of orthodontics
• /
• v.51 no.2
• /
• pp.77-85
• /
• 2021

Objective: To evaluate the accuracy of a multi-stage convolutional neural network (CNN) model-based automated identification system for posteroanterior (PA) cephalometric landmarks. Methods: The multi-stage CNN model was implemented with a personal computer. A total of 430 PA-cephalograms synthesized from cone-beam computed tomography scans (CBCT-PA) were selected as samples. Twenty-three landmarks used for Tweemac analysis were manually identified on all CBCT-PA images by a single examiner. Intra-examiner reproducibility was confirmed by repeating the identification on 85 randomly selected images, which were subsequently set as test data, with a two-week interval before training. For initial learning stage of the multi-stage CNN model, the data from 345 of 430 CBCT-PA images were used, after which the multi-stage CNN model was tested with previous 85 images. The first manual identification on these 85 images was set as a truth ground. The mean radial error (MRE) and successful detection rate (SDR) were calculated to evaluate the errors in manual identification and artificial intelligence (AI) prediction. Results: The AI showed an average MRE of 2.23 ± 2.02 mm with an SDR of 60.88% for errors of 2 mm or lower. However, in a comparison of the repetitive task, the AI predicted landmarks at the same position, while the MRE for the repeated manual identification was 1.31 ± 0.94 mm. Conclusions: Automated identification for CBCT-synthesized PA cephalometric landmarks did not sufficiently achieve the clinically favorable error range of less than 2 mm. However, AI landmark identification on PA cephalograms showed better consistency than manual identification.

• The korean journal of orthodontics
• /
• v.47 no.1
• /
• pp.50-58
• /
• 2017

Introduction: Identifying menton (Me) on posteroanterior cephalograms and three-dimensional (3D) cone-beam computed tomography (CBCT) images is difficult, because the midpoint of the symphyseal area is not identifiable after the mandibular symphysis fuses at an early age. The aim of this study was to evaluate the reliability of the identification of the genial tubercle (GT) in patients with mandibular asymmetry and to compare it with that of the traditional landmark, Me. Methods: The samples comprised 20 CBCT images of adults with mandibular asymmetry. Two examiners performed the identifications and measurements. Me and GT were marked, and the anteroposterior, vertical, and transverse distances to the three reference planes were measured on 3D-reconstructed CBCT images. The intra- and inter-examiner reliability of landmark identification of Me and GT were assessed using the intraclass correlation coefficient (ICC) and Bland-Altman plots. Results: The Me and GT landmarks showed excellent reliability ($ICC{\geq}0.993$) three-dimensionally. In the transverse evaluation, the ICC values of the GT (range, 0.997-0.999) tended to be slightly higher than those of Me (range, 0.993-0.996). In the Bland-Altman plots for the two separate assessments, Me showed a maximum error of 1.76 mm in the transverse direction, whereas the GT showed a maximum error of 0.96 mm in the 95% limit. Conclusions: Our results suggest that both Me and GT are clinically reliable and equally useful landmarks for the evaluation of mandibular asymmetry on CBCT images.

• Imaging Science in Dentistry
• /
• v.51 no.3
• /
• pp.299-306
• /
• 2021

Purpose: This study aimed to propose a fully automatic landmark identification model based on a deep learning algorithm using real clinical data and to verify its accuracy considering inter-examiner variability. Materials and Methods: In total, 950 lateral cephalometric images from Yonsei Dental Hospital were used. Two calibrated examiners manually identified the 13 most important landmarks to set as references. The proposed deep learning model has a 2-step structure-a region of interest machine and a detection machine-each consisting of 8 convolution layers, 5 pooling layers, and 2 fully connected layers. The distance errors of detection between 2 examiners were used as a clinically acceptable range for performance evaluation. Results: The 13 landmarks were automatically detected using the proposed model. Inter-examiner agreement for all landmarks indicated excellent reliability based on the 95% confidence interval. The average clinically acceptable range for all 13 landmarks was 1.24 mm. The mean radial error between the reference values assigned by 1 expert and the proposed model was 1.84 mm, exhibiting a successful detection rate of 36.1%. The A-point, the incisal tip of the maxillary and mandibular incisors, and ANS showed lower mean radial error than the calibrated expert variability. Conclusion: This experiment demonstrated that the proposed deep learning model can perform fully automatic identification of cephalometric landmarks and achieve better results than examiners for some landmarks. It is meaningful to consider between-examiner variability for clinical applicability when evaluating the performance of deep learning methods in cephalometric landmark identification.

• The korean journal of orthodontics
• /
• v.52 no.4
• /
• pp.287-297
• /
• 2022

Objective: To investigate the pattern of accuracy change in artificial intelligence-assisted landmark identification (LI) using a convolutional neural network (CNN) algorithm in serial lateral cephalograms (Lat-cephs) of Class III (C-III) patients who underwent two-jaw orthognathic surgery. Methods: A total of 3,188 Lat-cephs of C-III patients were allocated into the training and validation sets (3,004 Lat-cephs of 751 patients) and test set (184 Lat-cephs of 46 patients; subdivided into the genioplasty and non-genioplasty groups, n = 23 per group) for LI. Each C-III patient in the test set had four Lat-cephs: initial (T0), pre-surgery (T1, presence of orthodontic brackets [OBs]), post-surgery (T2, presence of OBs and surgical plates and screws [S-PS]), and debonding (T3, presence of S-PS and fixed retainers [FR]). After mean errors of 20 landmarks between human gold standard and the CNN model were calculated, statistical analysis was performed. Results: The total mean error was 1.17 mm without significant difference among the four time-points (T0, 1.20 mm; T1, 1.14 mm; T2, 1.18 mm; T3, 1.15 mm). In comparison of two time-points ([T0, T1] vs. [T2, T3]), ANS, A point, and B point showed an increase in error (p < 0.01, 0.05, 0.01, respectively), while Mx6D and Md6D showeda decrease in error (all p < 0.01). No difference in errors existed at B point, Pogonion, Menton, Md1C, and Md1R between the genioplasty and non-genioplasty groups. Conclusions: The CNN model can be used for LI in serial Lat-cephs despite the presence of OB, S-PS, FR, genioplasty, and bone remodeling.

• The korean journal of orthodontics
• /
• v.38 no.6
• /
• pp.427-436
• /
• 2008

Objective: The purpose of this study is to compare landmark position between cephalometric radiography and midsagittal plane projected images from 3 dimensional (3D) CT. Methods: Cephalometric radiographs and CT scans were taken from 20 patients for treatment of mandibular prognathism. After selection of land-marks, CT images were projected to the midsagittal plane and magnified to 110% according to the magnifying power of radiographs. These 2 images were superimposed with frontal and occipital bone. Common coordinate system was established on the base of FH plane. The coordinate value of each landmark was compared by paired t test and mean and standard deviation of difference was calculated. Results: The difference was from $-0.14{\pm}0.65$ to $-2.12{\pm}2.89\;mm$ in X axis, from $0.34{\pm}0.78$ to $-2.36{\pm}2.55\;mm$ ($6.79{\pm}3.04\;mm$) in Y axis. There was no significant difference only 9 in X axis, and 7 in Y axis out of 20 landmarks. This might be caused by error from the difference of head positioning, by masking the subtle end structures, identification error from the superimposition and error from the different definition.

• The korean journal of orthodontics
• /
• v.54 no.1
• /
• pp.48-58
• /
• 2024

Objective: To quantify the effects of midline-related landmark identification on midline deviation measurements in posteroanterior (PA) cephalograms using a cascaded convolutional neural network (CNN). Methods: A total of 2,903 PA cephalogram images obtained from 9 university hospitals were divided into training, internal validation, and test sets (n = 2,150, 376, and 377). As the gold standard, 2 orthodontic professors marked the bilateral landmarks, including the frontozygomatic suture point and latero-orbitale (LO), and the midline landmarks, including the crista galli, anterior nasal spine (ANS), upper dental midpoint (UDM), lower dental midpoint (LDM), and menton (Me). For the test, Examiner-1 and Examiner-2 (3-year and 1-year orthodontic residents) and the Cascaded-CNN models marked the landmarks. After point-to-point errors of landmark identification, the successful detection rate (SDR) and distance and direction of the midline landmark deviation from the midsagittal line (ANS-mid, UDM-mid, LDM-mid, and Me-mid) were measured, and statistical analysis was performed. Results: The cascaded-CNN algorithm showed a clinically acceptable level of point-to-point error (1.26 mm vs. 1.57 mm in Examiner-1 and 1.75 mm in Examiner-2). The average SDR within the 2 mm range was 83.2%, with high accuracy at the LO (right, 96.9%; left, 97.1%), and UDM (96.9%). The absolute measurement errors were less than 1 mm for ANS-mid, UDM-mid, and LDM-mid compared with the gold standard. Conclusions: The cascaded-CNN model may be considered an effective tool for the auto-identification of midline landmarks and quantification of midline deviation in PA cephalograms of adult patients, regardless of variations in the image acquisition method.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• v.36 no.4
• /
• pp.262-269
• /
• 2010

Introduction: Accurate diagnosis and treatment planning are very important for orthognathic surgery. A small error in diagnosis can cause postoperative functional and esthetic problems. Pre-existing 2-dimensional (D) chephalogram analysis has a high likelihood of error due to its intrinsic and extrinsic problems. A cephalogram can also be inaccurate due to the limited anatomic points, superimposition of the image, and the considerable time and effort required. Recently, an improvement in technology and popularization of computed tomography (CT) provides patients with 3-D computer based cephalometric analysis, which complements traditional analysis in many ways. However, the results are affected by the experience and the subject of the investigator. Materials and Methods: The effects of the sources human error in 2-D cephalogram analysis and 3-D computerized tomography cephalometric analysis were compared using Simplant CMF program. From 2008 Jan to 2009 June, patients who had undergone CT, cephalo AP, lat were investigated. Results: 1. In the 3 D and 2 D images, 10 out of 93 variables (10.4%) and 11 out 44 variables (25%), respectively, showed a significant difference. 2. Landmarks that showed a significant difference in the 2 D image were the points frequently superimposed anatomically. 3. Go Po Orb landmarks, which showed a significant difference in the 3 D images, were found to be the artificial points for analysis in the 2 D image, and in the current definition, these points cannot be used for reproducibility in the 3 D image. Conclusion: Generally, 3-D CT images provide more precise identification of the traditional cephalometric landmark. Greater variability of certain landmarks in the mediolateral direction is probably related to the inadequate definition of the landmarks in the third dimension.