Intraobserver reliability of landmark identification in cone-beam computed tomography-synthesized two-dimensional cephalograms versus conventional cephalometric radiography: A preliminary study

Accuracy of conventional versus cone-beam CT-synthesised lateral cephalograms for cephalometric analysis: A systematic review

OBJECTIVE

To assess the accuracy of cone-beam computed tomography (CBCT)-synthesised lateral cephalograms (CSLCs) compared with conventional lateral cephalograms for cephalometric analysis in human participants and skull models.

METHODS

The authors performed a search of PubMed, Scopus, Google Scholar and Embase databases on 4 October 2021. Included studies met the following criteria: published in English; compared conventional lateral cephalograms and CSLCs; assessed hard- and soft-tissue landmarks; and were performed on human or skull models. Data extraction from eligible studies was performed by two independent reviewers. The quality of evidence was assessed using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist tool - diagnostic accuracy studies.

RESULTS

A total of 20 eligible articles were included in this systematic review. Of these 20 studies, 17 presented with a low risk of bias, while three were found to have a moderate risk of bias. Hard- and soft-tissue analyses were evaluated for each imaging modality. The findings reveal that CSLCs are accurate and comparable to conventional lateral cephalograms for cephalometric analysis and demonstrate good inter-observer reliability. Four studies reported a higher accuracy with CSLCs.

CONCLUSION

Overall, the diagnostic accuracy and reproducibility of CSLCs were comparable to conventional lateral cephalograms in cephalometric analysis. It is justified that patients who have an existing CBCT scan do not need an additional lateral cephalogram, minimising unnecessary radiation exposure, expenses and time for the patient. Larger voxel sizes and low-dose CBCT protocols can be considered to minimise radiation exposure.

REGISTRATION

This study was registered with PROSPERO (CRD42021282019).

Cephalogram synthesis and landmark detection in dental cone-beam CT systems

Due to the lack of a standardized 3D cephalometric analysis methodology, 2D cephalograms synthesized from 3D cone-beam computed tomography (CBCT) volumes are widely used for cephalometric analysis in dental CBCT systems. However, compared with conventional X-ray film based cephalograms, such synthetic cephalograms lack image contrast and resolution, which impairs cephalometric landmark identification. In addition, the increased radiation dose applied to acquire the scan for 3D reconstruction causes potential health risks. In this work, we propose a sigmoid-based intensity transform that uses the nonlinear optical property of X-ray films to increase image contrast of synthetic cephalograms from 3D volumes. To improve image resolution, super resolution deep learning techniques are investigated. For low dose purpose, the pixel-to-pixel generative adversarial network (pix2pixGAN) is proposed for 2D cephalogram synthesis directly from two cone-beam projections. For landmark detection in the synthetic cephalograms, an efficient automatic landmark detection method using the combination of LeNet-5 and ResNet50 is proposed. Our experiments demonstrate the efficacy of pix2pixGAN in 2D cephalogram synthesis, achieving an average peak signal-to-noise ratio (PSNR) value of 33.8 with reference to the cephalograms synthesized from 3D CBCT volumes. Pix2pixGAN also achieves the best performance in super resolution, achieving an average PSNR value of 32.5 without the introduction of checkerboard or jagging artifacts. Our proposed automatic landmark detection method achieves 86.7% successful detection rate in the 2 mm clinical acceptable range on the ISBI Test1 data, which is comparable to the state-of-the-art methods. The method trained on conventional cephalograms can be directly applied to landmark detection in the synthetic cephalograms, achieving 93.0% and 80.7% successful detection rate in 4 mm precision range for synthetic cephalograms from 3D volumes and 2D projections, respectively.

Test-retest reliability of mandibular morphology measurements on cone-beam computed tomography-synthesized cephalograms with random head positioning errors

Background

Cephalometric radiography has been used for orthodontic and surgical treatment planning and assessment, and for quantifying mandibular growth. However, it remains unclear how head positioning errors and the level of examiner experience affect the reliability of such morphometric measurements. The current study aimed to bridge the gap by determining the intra-, inter-rater, and inter-session reliability of measurements of mandibular morphology with random head positioning errors as measured by a junior and a senior dentist.

Methods

Cone-beam computed tomography data of twelve mandibles were obtained with each rotated randomly away from the neutral position within the range of +3 and −3° along each of the anatomical axes to simulate six imaging trials. A synthetic cephalogram for each trial was obtained via a digitally reconstructed radiography (DRR) technique and eleven landmarks for twelve morphological parameters on the cephalogram were identified manually six times by a junior and a senior dentist. The procedure was repeated on another day within 5 days. Test–retest reliability was assessed in terms of an intra-class correlation coefficient (ICC) using a two-way mixed-effects model.

Results

Good to very good intra-rater (senior: ICC > 0.92; junior: ICC > 0.78), inter-rater (ICC > 0.70 for most parameters) and inter-session reliability (senior: ICC > 0.84; junior: ICC > 0.62) were found. Bland & Altman plots of inter-rater comparisons show that there were systematical biases between the examiners on most parameters, except for the distance between Gonion and Pogonion.

Conclusions

The current results suggest that good to very good intra-rater, inter-rater and inter-session reliability can be achieved for most parameters with randomized head positioning errors; higher inter-session reliability can be achieved by more experienced examiners; and that long-term monitoring of mandibular growth based on cephalographic measurements should be made by the same more experienced examiner. The current DRR-based approach can be used to evaluate individual factors that affect the morphological measurements.

Effect of changing the head position on accuracy of transverse measurements of the maxillofacial region made on cone beam computed tomography and conventional posterior-anterior cephalograms

OBJECTIVES

This study aimed to assess the effect of head position on the accuracy of transverse measurements of the maxillofacial region on CBCT and conventional posteroanterior (PA) cephalograms. The second objective of this study was to find skull positions with the greatest and smallest effect on transverse measurements in the maxillofacial region.

METHODS

PA cephalograms and CBCT scans were obtained from 10 dry human skulls in 7 positions, namely the central position, 10° and 20° rotations, 10° and 20° tilts and 10° and 20° tips. The CBCT scans were converted to PA cephalograms on which distances from six landmarks, namely the nasal cavity, zygomatic arch, jugale, antegonion, condylion and zygomaticofrontal suture to the mid-sagittal plane, were measured on both sides using Dolphin two-dimensional software. The paired t-test was used to compare the mean values separately in each position (for each landmark) with the gold standard (central skull position). The interclass correlation coefficient and the Bland-Altman plot were used to compare the mean values measured by two observers.

RESULTS

The mean values of the distances measured on CBCT PA cephalograms were greater than those measured on conventional PA cephalograms; this difference was statistically significant for some landmarks (p < 0.005). The rotated position (as compared with the central position) caused the greatest change in values for most landmarks on both sides (p < 0.005).

CONCLUSIONS

The CBCT PA cephalogram was more accurate than the conventional PA cephalogram, and landmarks farther from the midline exhibited greater changes on cephalograms compared with those closer to the midline. Patients are at risk of improper positioning when undergoing extraoral radiography such as PA cephalograms. Changes in head position may affect the transverse measurements and thus the treatment plan.

Comparison of measurements of mandible growth using cone beam computed tomography and its synthesized cephalograms

BACKGROUND

The current study aimed to compare the measurements of the mandible morphology using 3D cone beam computed tomography (CBCT) images with those using 2D CBCT-synthesized cephalograms; to quantify errors in measurements based on 2D synthesized cephalograms; and to clarify the effects such errors have on the description of the mandibular growth.

METHODS

Mandibles of six miniature pigs were scanned monthly using CBCT over 12 months and the data were used to reconstruct the 3D bone models. Five anatomical landmarks were identified on each bone model, and the inter-marker distances and monthly distance changes were calculated and taken as the gold standard. Synthetic 2D cephalograms were also generated for each bone model using a digitally reconstructed radiography (DRR)-generation method. Errors in cephalogram measurements were determined as the differences between the calculated variables in cephalograms and the gold standard. The variations between cephalograms and the gold standard were also compared using paired t-tests.

RESULTS

While the inter-marker distance increases varied among the marker pairs, all marker pairs increased their inter-marker distances gradually every month, reaching 50% of the total annual increases during the fourth and fifth months, and then slowing down in the subsequent months. The 2D measurements significantly underestimated most of the inter-marker distances throughout the monitoring period, in most of the monthly inter-marker distance changes during the first four months, and in the total growth (p < 0.05).

CONCLUSIONS

Significant errors exist in the measurements using 2D synthesized cephalogram, underestimating the mandibular dimensions and their monthly changes in the early stages of growth, as well as the total annual growth. These results should be considered in dental treatment planning at the beginning of the treatment in order to control more precisely the treatment process and outcome.