Reliability of landmark identification in cephalometric radiography acquired by a storage phosphor imaging system

Can artificial intelligence-driven cephalometric analysis replace manual tracing? A systematic review and meta-analysis

OBJECTIVES

This systematic review and meta-analysis aimed to investigate the accuracy and efficiency of artificial intelligence (AI)-driven automated landmark detection for cephalometric analysis on two-dimensional (2D) lateral cephalograms and three-dimensional (3D) cone-beam computed tomographic (CBCT) images.

SEARCH METHODS

An electronic search was conducted in the following databases: PubMed, Web of Science, Embase, and grey literature with search timeline extending up to January 2024.

SELECTION CRITERIA

Studies that employed AI for 2D or 3D cephalometric landmark detection were included.

DATA COLLECTION AND ANALYSIS

The selection of studies, data extraction, and quality assessment of the included studies were performed independently by two reviewers. The risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. A meta-analysis was conducted to evaluate the accuracy of the 2D landmarks identification based on both mean radial error and standard error.

RESULTS

Following the removal of duplicates, title and abstract screening, and full-text reading, 34 publications were selected. Amongst these, 27 studies evaluated the accuracy of AI-driven automated landmarking on 2D lateral cephalograms, while 7 studies involved 3D-CBCT images. A meta-analysis, based on the success detection rate of landmark placement on 2D images, revealed that the error was below the clinically acceptable threshold of 2 mm (1.39 mm; 95% confidence interval: 0.85-1.92 mm). For 3D images, meta-analysis could not be conducted due to significant heterogeneity amongst the study designs. However, qualitative synthesis indicated that the mean error of landmark detection on 3D images ranged from 1.0 to 5.8 mm. Both automated 2D and 3D landmarking proved to be time-efficient, taking less than 1 min. Most studies exhibited a high risk of bias in data selection (n = 27) and reference standard (n = 29).

CONCLUSION

The performance of AI-driven cephalometric landmark detection on both 2D cephalograms and 3D-CBCT images showed potential in terms of accuracy and time efficiency. However, the generalizability and robustness of these AI systems could benefit from further improvement.

REGISTRATION

PROSPERO: CRD42022328800.

Surface-based 3-dimensional cephalometry: An objective analysis of cranio-mandibular morphology

INTRODUCTION

The purpose of this study was to apply a previously reported homologous model for quantifying and visualizing the three-dimensional (3D) cranio-mandibular morphology.

METHODS

Twenty-eight 3D cone-beam computed tomography (CBCT) images of Japanese patients (7 to 13 years of age) showing skeletal Class 1 malocclusions were used in this study. Wire mesh fitting was conducted for each patient based on the assignment of landmarks to each 3D surface (homologous modeling). Cranial and mandibular images were generated using surface rendering of the CBCT images. The mean and standard deviation of each point on the wire mesh were calculated as the normative mean. Two operators conducted the process twice for seven patients, with a one-week interval between fittings. The intra- and inter-examiner reliability were determined using the 95% confidence interval minimal detectable change (MDC₉₅) for random error and intra-class correlation coefficients (ICCs).

RESULTS

The intra- and inter-examiner reliability of the mesh fitting method were almost perfect. Normative ranges of the cranial and mandibular surfaces of Japanese were determined.

CONCLUSIONS

A clinical method that enables practitioners to quantify and visualize the hard tissues of a patient's face in three dimensions with almost perfect reliability is presented. This method allows practitioners to evaluate how patients' skeletal characteristics differ from normative means in three dimensions.

Are orthodontic landmarks and variables in digital cephalometric radiography taken in fixed and natural head positions reliable?

OBJECTIVE

The purpose of the study was to evaluate the reliability of the most common landmarks and variables in digital lateral cephalometric radiographies in fixed and natural head positions.

MATERIALS AND METHODS

Twenty-one patients with anterior or distal displacement of the mandible treated in the Postgraduate Orthodontic Clinic of the Aristotle University of Thessaloniki had a digital lateral cephalometric radiography in fixed and in natural head position. The images where digitized. The main investigator and 6 examiners, digitized 61 landmarks and analysed 34 variables. We examined the intra-observer and inter-observer variability.

RESULTS

There was a significant difference in X-axis in the distribution (hence the mean results) according to the reliability of the landmarks and in only 2 variables in digital lateral cephalograms between fixed and natural head position.

CONCLUSIONS

Cephalometric landmarks and variables showed reliability in digital lateral cephalometric radiography in fixed and natural head position. In lateral cephalograms taken in fixed head position an anterior inclination of the head was noticed compared to those in natural head position.

Cephalometric landmark variability among orthodontists and dentomaxillofacial radiologists: a comparative study

Purpose

The aim this study was to compare the accuracy of orthodontists and dentomaxillofacial radiologists in identifying 17 commonly used cephalometric landmarks, and to determine the extent of variability associated with each of those landmarks.

Materials and Methods

Twenty digital lateral cephalometric radiographs were evaluated by two groups of dental specialists, and 17 cephalometric landmarks were identified. The x and y coordinates of each landmark were recorded. The mean value for each landmark was considered the best estimate and used as the standard. Variation in measurements of the distance between landmarks and measurements of the angles associated with certain landmarks was also assessed by a subset of two observers, and intraobserver and interobserver agreement were evaluated.

Results

Intraclass correlation coefficients were excellent for intraobserver agreement, but only good for interobserver agreement. The least reliable landmark for orthodontists was the gnathion (Gn) point (standard deviation [SD], 5.92 mm), while the orbitale (Or) was the least reliable landmark (SD, 4.41 mm) for dentomaxillofacial radiologists. Furthermore, the condylion (Co)-Gn plane was the least consistent (SD, 4.43 mm).

Conclusion

We established that some landmarks were not as reproducible as others, both horizontally and vertically. The most consistently identified landmark in both groups was the lower incisor border, while the least reliable points were Co, Gn, Or, and the anterior nasal spine. Overall, a lower level of reproducibility in the identification of cephalometric landmarks was observed among orthodontists.

The Effect of Emboss Enhancement on Reliability of Landmark Identification in Digital Lateral Cephalometric Images

Background:

Evaluation of the craniofacial bones is the oldest method to measure the facial proportion ratio in orthodontics.

Objectives:

The purpose of this study was to evaluate the effect of emboss enhancement on the reliability of landmark identification in digital lateral cephalometric images.

Materials and Methods:

Ten digital lateral cephalograms were selected from the archive of an oral and maxillofacial radiology center. Using DIGORA software, these images were saved in two formats; common images and 3D emboss images. On these images, 32 skeletal, dental, and soft tissue landmarks were marked at least twice with a 2-week interval by four observers (two radiologists and two orthodontists). In order to determine the position of the marked landmarks (in x and y coordinates), a software was designed. The statistical analysis was performed in SPSS software and the reliability of each observer was obtained by means of intraclass correlation coefficient (ICC).

Results:

In three skeletal landmarks [Orbit (Or), condyl top (Cond), and pogonion (Pog)], the enhancement caused significant reduction in the reliability, and in four skeletal [Anterior Nasal Spine (ANS), B, A, and Basion (Ba)], two dental (U1 root, L1 incisal), and one soft tissue landmark (Menton soft tissue), the enhancement increased the reliability of landmark detection between the two phases of the study. Totally, ICC of embossed images in both x and y coordinates were greater than the typical images, but the difference was not statistically significant. However, the effect of enhancement on the improvement of the reliability of landmark identification was higher in the x-axis than the y-axis.

Conclusions:

Using embossed images is only effective in increasing the reliability of detection in a few numbers of cephalometric landmarks.

Validation of conventional 2D lateral cephalometry using 3D cone beam CT

OBJECTIVE

To determine if two-dimensional (2D) measurements from conventional cephalometric lateral skull radiographs are comparable to those derived from three-dimensional (3D) cone beam computed tomography (CBCT) images.

DESIGN

In vitro laboratory study.

SETTING

University Dental Hospital.

METHODS

A sample size calculation determined that 14 dried skulls were required to detect a 2° difference in angular measurements. The skulls were scanned at 0·3 mm(3) voxel size. Maximum intensity projection (MIP) views were uploaded into OPAL cephalometric software (British Orthodontic Society, London, UK) for 2D analysis. CBCT data was uploaded into Mimics (Materialise, Leuven, Belgium) with 3D reconstructed and sagittal slice views being used. An Eastman analysis was carried out for the 2D and 3D images with the data compared using two-sample t-tests at P<0·05. Measurements greater than 2° between the 2D and 3D data were considered clinically significantly different. Intra-observer reproducibility was assessed by calculating random and systematic error using the Dahlberg formula and a two-sample t-test (P<0·05).

RESULTS

The random error was below 0·5° and the systematic error was acceptable (P<0·05). There were no statistically significant differences between the measurements from the 2D and 3D images for any variable (P<0·05). However, the mean SNB value and the mean value for the angle between the lower incisor and mandibular plane differed by greater than 2° between the 2D and 3D data. The latter was thought to be due to limitations of the definition of Gonion for 3D images and the precision of locating the lower incisor apex in 2D.

CONCLUSION

Measurements used in the Eastman cephalometric analysis derived from 2D cephalometric lateral skull images are comparable to those derived from 3D CBCT images.

Evaluation of the reliability of measurements in cephalograms generated from cone beam computed tomography

OBJECTIVE: The purpose was to compare angular and linear measurements generated in digital cephalometric radiographs and cephalograms synthesized from three-dimensional images. METHODS: Twenty-six individuals (12 men and 14 women) with mean age of 26.3 years were selected. Digital cephalometric radiographs and CBCTs were taken on the same day. The images were imported and analyzed on Dolphin Imaging V.10.5 software, which synthesized cephalograms in perspective projection and magnification of 9.7%. A single observer marked the points and repeated the procedure with an interval of time of ten days to evaluate intraexaminer error. In the statistical analysis paired Student's t test was used to establish the correlation between the measurements. RESULTS: The angular measurements GoGn.SN and IMPA, which involved the Gonial point (Go) and the linear measurements that involved the lips presented significant difference (p < 0.05). The other measurements presented good correlation. CONCLUSION: The measurements in the synthesized cephalograms proved to be reliable.

Landmark identification errors on cone-beam computed tomography-derived cephalograms and conventional digital cephalograms

INTRODUCTION

In this study, we investigated the landmark identification errors on cone-beam computed tomography (CBCT)-derived cephalograms and conventional digital cephalograms.

METHODS

Twenty patients who had both a CBCT-derived cephalogram and a conventional digital cephalogram were recruited. Twenty commonly used lateral cephalometric landmarks and 2 fiducial points were identified on each cephalogram by 11 observers at 2 time points. The mean positions of the landmarks identified by all observers were used as the best estimate to calculate the landmark identification errors. In addition to univariate analysis, regression analysis of landmark identification errors was conducted for identifying the predicting variables of the observed landmark identification errors. To properly handle the multilayer correlations among the clustered observations, a marginal multiple linear regression model was fitted to our correlated data by using the well-known generalized estimating equations method. In addition to image modality, many variables potentially affecting landmark identification errors were considered, including location and characteristics of the landmark, seniority of the observer, and patient information (sex, age, metallic dental restorations, and facial asymmetry).

RESULTS

Image modality was not the significant variable in the final generalized estimating equations model. The regression coefficient estimates of the significant landmarks for the overall identification error ranged from -0.99 (Or) to 1.42 mm (Ba). The difficulty of identifying landmarks on structural images with multiple overlapping--eg, Or, U1R, L1R, Po, Ba, UMo, and LMo--increased the identification error by 1.17 mm. In the CBCT modality, the identification errors significantly decreased at Ba (-0.76 mm).

CONCLUSIONS

The overall landmark identification errors on CBCT-derived cephalograms were comparable to those on conventional digital cephalograms, and Ba was more reliable on CBCT-derived cephalograms.

Evaluation of the CT scanogram for assessment of craniofacial morphology

OBJECTIVE

The purpose of this prospective laboratory study was to investigate the suitability of the computed tomography scanogram (CT) as compared with conventional (CC) or storage phosphor digital (SP) lateral cephalograms.

MATERIALS AND METHODS

Twenty intact, adult-sized, dry human skulls were used. Each skull was fixed in a custom-made plastic box and imaged in each of the three radiographic machines. All 60 hard copy radiographs were directly digitized using a customized cephalometric program, and all skulls were measured with an external caliper. All measurements were carried out twice, 2 weeks apart. "A priori" ranges were defined in advance for all 18 craniofacial parameters. Linear measurements were corrected for radiographic magnification. Data were analyzed using coefficients of repeatability, limits of agreement, and paired t-tests.

RESULTS

Direct skull, anterior face height, sella-nasion, and cephalometric SnMx, SNA, SNB, and ANB (except CT) measurements were repeatable. Sella-basion, basion-nasion, most posterior face height measurements, MxMn, and incisor angulations were either not repeatable or not in agreement. When statistical significance was noted, however, the mean difference was of small magnitude for nearly all parameters.

CONCLUSION

Depending on the clinical situation and the degree of precision required, the scanogram is a viable alternative to lateral cephalometry for assessment of sella-nasion, anterior face height, skeletal planes angles (SnMx, MxMn), SNA, SNB, and ANB, but not of sella-basion, basion-nasion, posterior face height measurements, and incisor angulations.

Reliability and the smallest detectable differences of lateral cephalometric measurements

INTRODUCTION

The aim of this study was to determine the reliability and the measuring error (by means of the smallest detectable error) of 11 angular and 4 linear measurements commonly used for cephalometric analysis.

METHODS

Twenty-five digital lateral cephalograms were randomly selected and traced with Viewbox software (version 3.1.1.13, dHAL Software, Kifissia, Greece). This was repeated 3 times by 2 observers during 3 sessions. There was at least 1 week between each session. Differences were analyzed with a repeated measurement analysis of variance (ANOVA). Intraobserver and interobserver reliabilities were calculated with intraclass correlation coefficients (ICC) based on absolute agreement. Measurement error was determined by means of the smallest detectable difference.

RESULTS

The intraobserver agreement of the measurements was good (ICC >0.82). SNA, SNB, ANB, and ANS-Me had the smallest intraobserver errors for both observers (>1.86 mm or degrees). Except for SN-FH (ICC = 0.76), interobserver agreement was good (ICC >0.87).

CONCLUSIONS

Determining the appropriate measuring error of cephalometric measurements by means of the smallest detectable difference is necessary to find the true difference between the start and the end of active treatment. Depending on the magnitude of clinical significance, the measuring error was possibly clinically significant for all variables tested and, therefore, questions the use of these variables to detect the true treatment effect.

Influence of a programme of professional calibration in the variability of landmark identification using cone beam computed tomography-synthesized and conventional radiographic cephalograms

OBJECTIVES

The validity of any measurement obtained through a cephalogram largely depends on the reproducibility of the cephalometric landmarks. The purpose of this study is to evaluate the influence of a programme of professional calibration (PPC) on the variability of landmark identification comparing conventional radiographs and cone beam CT (CBCT)-synthesized cephalograms.

METHODS

5 graduate students in oral radiology identified 20 cephalometric landmarks from cephalograms generated from conventional radiographs (RADs), Ray-Sum CBCT-synthesized cephalograms (CBTs) and half-skull CBT (HSTs) from 10 patients. After a period of reinforcement on instruction and calibration with inter- and intraexaminer assessment of reproducibility (intraclass coefficient correlation scores > 0.75) for RADs, CBTs and HSTs obtained from 5 different patients, observers were asked to repeat the analysis of the first 10 patients under the same circumstances. Values in millimetres represented each landmark in a table of Cartesian co-ordinates (x- and y-axes).

RESULTS

ANOVA showed significant reduction in variability levels after the PPC, and there were no differences among the methods of image acquisition. Repeated measures ANOVA indicated that the PPC accounted for reduction in variability levels in 14 of 20 landmarks.

CONCLUSIONS

The results suggest that a PPC has more influence than the type of image acquisition on variability of landmark identification based on two-dimensional cephalometric analysis. Cephalograms obtained from RAD or CBCT can be considered equivalent for clinical and experimental applications.

Avaliação de métodos indiretos de digitalização de radiografias cefalométricas em comparação ao método digital direto

OBJETIVO: avaliar métodos de digitalização indireta de radiografias cefalométricas em comparação ao método digital direto. MÉTODOS: a amostra foi composta de dez radiografias cefalométricas adquiridas pelo Orthopantomograph OP100/Orthocef OC100 (GE - Instrumentarium), digital direto. As imagens foram inseridas no programa Adobe Photoshop® e cinco pontos cefalométricos foram marcados. Procedeu-se a impressão em transparência e foi feita a digitalização indireta das imagens impressas, por meio das câmeras fotográficas digitais Sony® DSC-W5 e Canon® Rebel XT/EOS 350D, fixas em estativa, nas distâncias de 25cm e 60cm e com o scanner Scan Jet 4C Hewlett Packard®. As imagens diretas e indiretas foram inseridas e calibradas no software Radiocef Studio (Radiomemory®, Brasil), marcando-se o centro dos pontos previamente assinalados. A análise cefalométrica computadorizada gerou três grandezas angulares e quatro lineares, as quais foram submetidas a tratamento estatístico. RESULTADOS: as imagens do scanner Scan Jet 4C HP® demonstraram pequenas alterações significativas, porém, sem relevância clínica. Quando da digitalização a 60cm da radiografia, as duas câmeras causaram distorções que foram significativas mas clinicamente aceitáveis; e, a 25cm, causaram as maiores distorções, com importância clínica, na Canon® Rebel XT. CONCLUSÃO: o scanner Scan Jet 4C HP® com leitor de transparências foi o melhor método e as câmeras operando a 60cm mostraram-se adequadas para digitalização de radiografias. As câmeras na distância de 25cm causaram distorções na imagem que alteraram as medidas lineares, podendo a Canon® Rebel XT comprometer o diagnóstico ortodôntico.

Intraexaminer and interexaminer reliabilities of landmark identification on digitized lateral cephalograms and formatted 3-dimensional cone-beam computerized tomography images

INTRODUCTION

The purposes of this study were to determine and compare the intraexaminer and interexaminer reliabilities of commonly used cephalometric landmarks identified on digitized lateral cephalograms and formatted cone-beam computerized tomography (CBCT) images.

METHODS

CBCT images from 10 randomly selected adolescent patients were obtained from the orthodontic records of a private practice. Measurement errors, and intraexaminer, and interexaminer reliability correlation coefficients (ICC) were obtained for all landmark coordinates.

RESULTS

Intraexaminer and interexaminer reliabilities for all coordinates for most landmarks on the digital lateral cephalograms and CBCT images were greater than 0.9 (ICC value). The means of landmark locations differed by approximately 1 mm in most coordinates from the lateral cephalograms and were predominantly higher than 1 mm for all coordinates from the CBCT images.

CONCLUSIONS

Intraexaminer and interexaminer reliabilities were high for most landmarks. Coordinates with greater measurement errors in the lateral cephalograms (condylion, gonion, porion, mandibular incisor apex, and posterior nasal spine) were in structures without clearly defined borders. In the CBCT images, gonion, condylion, and porion were located on surfaces that were flat or curved, making it difficult to recognize a specific reference point. Other less reliable landmarks (anterior nasal spine, posterior nasal spine, mandibular incisor apex) were located in structures with lower densities and could not be visualized with 3-dimensional reconstruction; thus, they had high measurement errors.

Image distortion and magnification of 3 digital CCD cephalometric systems

OBJECTIVE

The magnification and distortion of images made from scanning and nonscanning CCD cephalometric systems was evaluated.

STUDY DESIGN

Acrylic box and dry human skull phantoms were imaged using a conventional cephalostat, the nonscanning Kodak 8000C, the horizontally scanning GE/Instrumentarium OC100D, and the vertically scanning Sirona OrthophosDS. True linear and angular measurements of the phantoms were made using a coordinate measuring system and the Hitachi MercuRay cone beam system. The accuracy of linear and angular measurements was assessed, as were magnification and distortion, where appropriate.

RESULTS

Statistically significant differences between linear and angular measurements were found for almost all measurements for both phantoms. In general, the conventional cephalostat and Sirona OrthophosDS systems produced the greatest magnification and distortion while the GE/Instrumentarium OC 100D and Kodak 8000C systems, the least.

CONCLUSIONS

Measurement differences related to the beam geometries of these systems could not be predicted a priori. Unaccounted for, these differences could result in clinically significant consequences.

Surgical cephalometrics: applications and developments

LEARNING OBJECTIVES

After studying this article, the participant should be able to: 1. Describe the historical origins of modern cephalometry. 2. Identify common landmark points on the lateral cephalogram. 3. Describe multiple common clinical uses for cephalometry. 4. Exhibit knowledge of developments in imaging and analysis alternatives.

BACKGROUND

Interest in the dimensions of the human head has been present since antiquity. Proportional analysis and measures from cadaveric specimens led to the development of radiologic image capture and analysis on living subjects. These techniques were originally applied to establishing normative values, documenting growth, and diagnosing dentofacial disharmonies. This article reviews the origins of cephalometric methodology and current developments and applications.

METHODS

The authors conducted a MEDLINE search and review of all English language articles using the keywords "cephalometric" and "cephalometrics."

RESULTS

Cephalometrics have undergone substantial use and development since the introduction of radiologic imaging on living human subjects in 1931. Although frequently associated with orthognathic surgery, cephalometrics have been applied to a number of conditions involving altered craniofacial morphology. Advances in imaging and computing have led to increased interest in three-dimensional and non-x-ray-based assessment of the human head. Mathematical models have been applied to standard cephalometric information to increase the descriptive accuracy of the complex shapes involved.

CONCLUSIONS

Cephalometric techniques and analyses are versatile tools that can be applied to a wide variety of clinical scenarios involving the craniofacial region. New technologies and expanded applications promise to continue the development and use of this well-established methodology.

Comparison of conventional and cone beam CT synthesized cephalograms

OBJECTIVES

To compare cephalometric measurements from synthesized cone beam CT (CBCT) lateral cephalograms using orthogonal and perspective projections with those from conventional cephalometric radiographs and dry skulls.

METHODS

Ten skulls were imaged using CBCT and conventional cephalometry. CBCT volume data were exported in DICOM format and imported in Dolphin 3D (pre-release version). Orthogonal and perspective lateral cephalometric radiographs were created from 3D virtual models. Nine linear and five angular measurements were made in Dolphin at three different times. Three calliper measures of midsagittal landmarks were made directly onto skulls. Perspective and conventional image measurements were corrected for known magnification. Reproducibility of measurements was assessed using multivariate analysis of variance (MANOVA). Linear and angular measurements were compared between image modalities by measurement using a repeated measures MANOVA model. Differences and absolute value of differences between image measurements and skull measurements were assessed using analysis of variance (ANOVA).

RESULTS

Measurements were not different between the imaging modalities (P>0.05), except for the mandibular unit length (P=0.01). Linear midsagittal measurements were significantly greater than skull measurements for perspective CBCT and significantly less than skull measurements for conventional images (P=0.003). Precision of orthogonal CBCT midsagittal linear measurements was significantly better than the other modalities (P=0.007). Orthogonal CBCT projections provided more accurate midsagittal skull measurements than perspective CBCT or conventional cephalometric radiographs.

CONCLUSIONS

CBCT can reproduce conventional cephalometric geometry with similar precision and accuracy. Orthogonal CBCT projections provided greater accuracy of measurement for midsagittal plane dimensions than perspective CBCT or conventional cephalometric images.