Introduction
This study aimed to analyze the image quality of standard dose orthopantomograms (sd-PAN), extracted PAN from standard dose cone-beam computed tomography (CBCT [extr-PAN]), and from reduced-dose ultra low dose-low dose CBCT (rd-PAN).
Methods
Image pairs, sd-CBCT and sd-PAN, of patients were selected if taken within 6 months of each other. From the sd-CBCT, an extr-PAN was extracted, and a simulated rd-PAN (sim rd-PAN) was constructed using a filtering technique. Three experienced orthodontists assessed image quality by means of 9 yes-and-no statements. For subjective statements (1-5), observers’ opinions of technical acceptability and visibility of anatomic structures, observers indicated their disagreement or agreement. For objective statements (6-9), regarding comparability to a gold standard, observers indicated the presence of dental structures. Positive response rates were calculated for the subjective statements. Agreement rates with the gold standard were calculated for the objective statements. Thresholds for acceptable image quality were if ≥2 observers agreeing with the subjective statements or agreeing with the gold standard (objective statements) for 90% for sd-PANs and 85% for both extr-PANs and sim rd-PANs.
Results
In total, images of 43 patients were included. For the subjective statements, only the sd-PAN met the threshold for technical adequacy (95.3%) and suitability for orthodontic treatment planning (95.3%). For objective statements 6, 7, and 9, extr-PAN and sim rd-PAN images met the threshold for acceptable quality, whereas the sd-PAN met the threshold for statements 6 and 9. Differences in agreement with the gold standard among images were small (85.7%-87.8%).
Conclusions
Sd-PAN images demonstrated superior image quality. Although extr-PAN and sim rd-PAN showed reduced diagnostic image quality for orthodontic diagnosis, differences among the 3 types of PAN regarding the gold standard were small.
Highlights
-
•
Ultra low dose cone beam computed tomography (CBCT) reduces radiation by 87% vs standard CBCT.
-
•
Could ultra low dose CBCT become the new standard in orthodontics?.
-
•
This study compares PAN image quality from ultra low and standard dose CBCT.
-
•
Findings show PAN from ultra low dose CBCT is usable in orthodontics.
-
•
A key benefit is 3-dimensional data from ultra low dose CBCT over standard PAN.
For orthodontic diagnosis, treatment planning, evaluation of treatment progress, and monitoring of growth and development, 2-dimensional (2D) lateral cephalograms (LC) and panoramic radiographs (PAN) have traditionally been used. These imaging techniques are favored for their ability to provide comprehensive views of the maxillofacial region and for their ease of use. However, cone-beam computed tomography (CBCT) has introduced new possibilities to maxillofacial imaging, offering 3-dimensional (3D) reconstructions and enhanced diagnostic capabilities by means of less overprojection, higher anatomic accuracy, 3D data integration, virtual 3D models, and predictive simulations. ,,
The advantages and disadvantages of standard dose CBCT (sd-CBCT) compared with the standard dose PAN (sd-PAN) and standard dose LC (sd-LC) have been widely discussed in the literature. ,, Unlike 2D radiography, CBCT provides volumetric, surface, and sectional information about craniofacial structures, eliminating issues such as magnification, distortion, and overprojection of anatomic structures. The additional third dimension may enhance orthodontic diagnosis and treatment planning. ,,,,, However, one considerable disadvantage of sd-CBCT is its 5-7 times higher radiation dose compared with the combined doses of an sd-PAN plus sd-LC. To address this issue, ultra-low dose (ULD) and ultra low dose-low dose (ULD-LD) CBCT protocols have been developed, achieving an 87% reduction in radiation dose compared with sd-CBCT protocols in both pediatric and adult phantoms , (for abbreviations see Table I ).
Table I
Abbreviations, source images, and extracted images
| CBCT | LC | PAN | |
|---|---|---|---|
| Standard dose | Standard dose | Standard dose | |
| sd-CBCT | sd-LC | sd-PAN | |
| Source image | Extracted image | Extracted image | Simulated PAN based on a filtered extr-PAN |
| sd-CBCT | extr-LC | extr-PAN | |
| ULD-LD-CBCT | rd-LC | rd-PAN | |
| extr-PAN | sim rd-PAN |
CBCT , cone beam computed tomography; LC , lateral cephalogram; PAN , orthopantomogram.
From sd-CBCT datasets, an LC and a PAN can be extracted, an extr-LC and extr-PAN respectively, offering a potential replacement for traditional 2D radiographs. Studies comparing cephalometric measurements performed on sd-LC with those on extr-LC found no significant differences, suggesting extr-LCs as viable replacements for sd-LCs. ,,,
Although sd-CBCT scans cannot yet fully replace PANs, they could eliminate the need for additional panoramic imaging if a recent sd-CBCT scan of both jaws is available. , Similarly, from a reduced-dose CBCT, also named ULD-LD CBCT, a reduced-dose LC and PAN can be extracted (rd-LC and rd-PAN). Orthodontic cephalometric measurements on rd-LC differed marginally from measurements on sd-LC, implying that rd-LC could be used for orthodontic purposes.
If the image quality of extracted rd-PAN is similar to that of sd-PAN, a single ULD-LD CBCT could become the standard imaging modality in orthodontics. This standard would not only provide 3D information but also contribute to a considerable reduction in radiation dose.
CBCT technology is expanding rapidly, with substantial growth in production and global adoption. The CBCT market is estimated to grow at an annual growth rate of 11.1% from $1.2 billion in 2022 to $2.27 billion in 2028. As CBCT units become more common, it is expected that in 2030, most dental practices in the United States will own a CBCT unit. In 3 United Kingdom pediatric dental practices, a rise in CBCT examinations was reported. Most of these examinations were undertaken for orthodontic purposes. This development is disturbing from a dosimetric point of view, because CBCT specifically poses concerns for pediatric patients because of their higher radiosensitivity and smaller body size, resulting in a higher effective dose compared with plain films. These factors increase the potential for adverse effects in children compared with adults. The mission of the “Image Gently” movement is to reduce the radiation dose in pediatric imaging.
There is limited evidence regarding the image quality of these reduced-dose radiographs. , Therefore, the current study aims to evaluate the image quality of sd-PAN, extr-PAN, and rd-PAN for orthodontic applications.
Material and methods
For this study, routinely collected sd-CBCT scans and sd-PANs from the existing database of the Departments of Orthodontics and Maxillofacial Surgery at the University Medical Center Groningen, the Netherlands, were used. The database contains x-ray images from 2013 to the present. First, all patients for whom both an sd-CBCT and an sd-PAN were available were selected (n = 270). The images were made for various indications determined by either the orthodontic or oral surgery department. In some instances, the PAN was made to assess the position of specific teeth in conjunction with surgical interventions. In other instances, the images were required for follow-up after orthodontic or surgical interventions. Occasionally, a CBCT was made in the oral surgery department and the PAN in the orthodontic department (or vice versa), for differing indications.
Image pairs were excluded if the time interval between the sd-PAN and sd-CBCT exceeded 6 months, if the images were incomplete (missing maxilla and/or mandible), or if images showed severe pathology or cleft lip and palate. Only PANs of good quality were included. After exclusion, image pairs of 43 patients remained. All images were anonymized.
The Institutional Medical Ethics Review Board judged that this study was not clinical research with human subjects as meant in the Medical Research Involving Human Subjects Act. Therefore, ethical approval was not required (No. 2022/394).
The sd-CBCT scans had been made using the KaVo 3D eXam CBCT unit (KaVo Dental GmbH, Bismarckring, Germany) for scans before April 2016, and using the Planmeca ProMax 3D Mid (Planmeca Oy, Helsinki, Finland) for scans after April 2016. The KaVo 3D eXam CBCT unit used a 170 × 230 mm field of view (FOV), set at 120 kV and 42.5 mAs with an isotropic voxel size of 0.3 mm. The Planmeca ProMax 3D used a 170 × 200-mm FOV, set at 90 kV and 20.3 mAs with an isotropic voxel size of 0.3 mm.
The PAN images had been made using the Planmeca ProMax 3D Mid (Planmeca Oy) with a 2780 × 1150 mm FOV, set at 64-70 kV and 70.0-150.0 mAs. This range of settings depended on the required protocols being used. Images taken before April 2016 were obtained using the Planmeca Promax (Planmeca Oy) under similar conditions and protocols.
From the sd-CBCT, an extr-PAN was extracted using the “superpan” function in Romexis software (version 6.4.3.33, Planmeca Oy). The “superpan” function automatically detects the patient’s arch and applies the image enhancement algorithm to improve the image quality. The “superpan” function involves using a wide focus band rather than a line. This band encompasses the entire bony structure of the maxilla and mandible. As a result, most structures within this band are depicted in a single PAN image. Figure 1 shows the sd-PAN ( Fig 1 , A ) and the extr-PAN ( Fig 1 , B ) of the same patient.
A , sd-PAN; B, extr-PAN; C, sim rd-PAN.
To simulate rd-PAN, a dry skull from an existing collection of skulls at the University Medical Center Groningen was placed in a 2 mm thick expanded polystyrene box to which a 1 cm thick layer of utility wax was applied to simulate soft tissues.
The first image, an sd-CBCT, was set at 90 kV and 109.9 mAs with an isotropic voxel size of 0.3 mm. The second image, a ULD-LD CBCT, was set at 90 kV and 15.7 mAs. Fields of view were identical.
From the sd-CBCT, an extr-PAN was extracted, and from the ULD-LD CBCT, an rd-PAN was extracted. To simulate rd-PANs (reduced-dose extr-PAN from ULD-LD CBCT), a software filter was designed and applied to all the extr-PANs from the patients in the database, resulting in sim rd-PANs. The design of this optic filter was based on the visual differences between an extr-PAN and an rd-PAN using Photoshop (version 2024; Adobe Systems Incorporated, San Jose, Calif). After enlargement, we compared the appearance of the 2 different images and adjusted the copy of the extr-PAN in Photoshop until it was visually identical to the rd-PAN. The adjustments consisted of resizing the images from 72 dots per inch (dpi) to 150 dpi. To add noise, a Gaussian blur ratio of 1 pixel was added. Finally, a Pixelate mosaic was added with a cell size of 3 square (3 × 3 pixels) to simulate a lower resolution. Figure 2 , A is an example of a true rd-PAN from a dry skull, whereas Figure 2 , B is the extr-PAN of the same skull with the filter to simulate the rd-PAN (sim rd-PAN), including an insert of a magnification of the same area of each image. A comparison of these two types of images ( Fig 2 , A and B ) shows the effect of the filter.
A , rd-PAN (skull); B, extr-PAN + filter (skull).
This optical filter ( Fig 3 ) was applied to all extr-PANs of the included patients to simulate rd-PANs, named sim rd-PANs ( Fig 1 , C ).
Simulation procedure.
Three orthodontists, each with >20 years of experience in interpreting orthodontic radiographs, assessed the 3 types of images, sd-PAN, extr-PAN, and sim rd-PAN, of the 43 patients. All images were presented in a random order in the original resolution against a black background, using a Microsoft 365 PowerPoint presentation (Microsoft, Redmond, Wash) on a Hewlett-Packard (HP) monitor (model HP 24i G4, resolution 1920 × 1200 dpi; HP, Palo Alto, Calif). Before the assessments, a calibration session was held. To determine intrarater reliability, sd-PAN, extr-PAN, and sim rd-PAN of 8 patients were presented twice in the show. The reduced ambient light conditions were similar to the normal radio-diagnostic situation.
Each image was accompanied by 9 statements, focusing on the diagnostic quality of anatomic structures. Answering options were yes or no. These statements were piloted on sd-PAN, extr-PAN, and sim rd-PAN of 10 patients, by a radiologist and a dental specialist, not further involved in the study. Their feedback was used to adapt the statements.
The following adapted statements were presented: (1) the image, as a whole, is technically acceptable and usable, (2) the image is suitable for assessing the course of the mandibular canal, (3) the image is suitable for assessing the condyles, (4) the image is suitable for assessing the course of the floor of the maxillary sinus, (5) an orthodontic treatment plan can be made based on this PAN, (6) the (germ of) third molar (tooth number) is present, (7) the root of tooth (tooth number) has been fully developed, (8) the roots of teeth (tooth number) and (tooth number) are parallel, and (9) fixed appliances (brackets) in the upper and/ or lower jaw are present.
Three orthodontists assessed the images independently under the same conditions. They were unable to scroll back within the show. The statements were presented through Google Forms (Google LLC, Mountain View, Calif, 2024).
For subjective statements 1-5, the assessor indicated whether they agreed or not with the statement. For objective statements 6-9, a gold standard was determined on the full FOV sd-CBCT dataset of the patients (RvB). For determining the gold standard, the relevant anatomic structures were examined using the scroll function in the Romexis software (version 6.4.3.33; Planmeca Oy). This method is accepted for creating a gold standard. ,,, Root development was assessed by determining whether the root was in Demirjian stage H. Root parallelism was assessed by drawing an imaginary plane perpendicular to the axis running through both teeth. To check for the presence of the third molar (or its germ) and the presence of fixed orthodontic appliances, the scroll function was used across different planes in the area of interest.
Statistical analysis
Intrarater reliability for the 3 observers was determined by calculating Cohen’s κ for the 24 images that were evaluated twice.
To analyze differences in scoring among observers, Cochran’s Q test for related samples was applied. A post-hoc pairwise comparison with Bonferroni correction for multiple tests was performed for the answers to statements 1-5 together and the sum of answers to statements 6-9 together.
For statements 1-5, the percentage of statements answered with “yes” was calculated for each observer. The percentage of patients for whom at least 2 observers deemed the image quality acceptable was calculated for statements 1-5 for each image type. The minimum threshold for acceptable image quality was set at 90% for sd-PAN and 85% for both extr-PAN and rd-PAN.
For statements 6-9, the percentage of agreement with the gold standard was calculated for each observer and image type. The percentage of patients for whom at least 2 observers agreed with the gold standard was calculated for statements 6-9 for each image type. The minimum threshold for acceptable image quality was set at 90% for sd-PAN and 85% for both extr-PAN and rd-PAN.
To analyze differences in percentages of acceptable image quality per image type, a Cochran’s Q test for related samples was applied, using Bonferroni correction for multiple tests.
To analyze differences in percentages of agreement with the gold standard per image type, a Cochran’s Q test for related samples was applied, using Bonferroni correction for multiple tests. All statistical analyses were performed using SPSS software (version 28; IBM, Armonk, NY).
Results
Intraobserver reliability expressed as Cohen’s κ was 0.824 for observer 1, 0.831 for observer 2, and 0.782 for observer 3.
Regarding statements 1-5, observer 1 rated 47.9% of the images acceptable, whereas observers 2 and 3 rated 52.6% and 33.8% of the images acceptable, respectively ( P <0.001; related samples Cochran’s Q test). Pairwise comparisons showed that differences between observer 3 and 1 and 3 and 2 were significant (both P <0.001).
Regarding statements 6-9, observer 1 agreed with the gold standard in 88.2% of the images, whereas observers 2 and 3 agreed with the gold standard in 86.8% and 84.7%, respectively ( P <0.05, related samples Cochran’s Q test). Pairwise comparisons showed that the difference between observer 3 and observer 1 was significant ( P = 0.028).
Regarding statements 1-5, positive scores were more frequent for sd-PAN images than extr-PAN and sim rd-PAN images ( Fig 4 ). Neither extr-PAN nor sim rd-PAN met the threshold for acceptable image quality for any of the statements ( Table II ). The sd-PAN images were rated technically acceptable and suitable for making an orthodontic treatment plan (statements 1 and 5).
Statements 1-5: percentages of statements answered with “yes”/ judged positively by observers 1, 2, and 3 per image type. Technically acceptable , the image, as a whole, is technically acceptable and usable; Mandibular canal , the image is suitable for assessing the course of the mandibular canal; Condyles , the image is suitable for assessing the condyles; Floor maxillary sinus , the image is suitable for assessing the course of the floor of the maxillary sinus; Treatment plan , an orthodontic treatment plan can be made based on this PAN. sd-PAN , standard dose PAN; extr-PAN , extracted PAN; sim rd-PAN , simulated extracted ultra low dose-low dose (ULD LD) PAN.
Table II
Statements 1-5: the percentage of images for which ≥2 observers judged image quality to be acceptable
| Statement | sd-PAN | extr-PAN | sim rd-PAN |
|---|---|---|---|
| Technically acceptable | 95.3 | 39.5 | 34.9 |
| Mandibular canal | 39.5 | 16.3 | 18.6 |
| Condyles | 76.7 | 9.3 | 9.3 |
| Floor maxillary sinus | 74.4 | 37.2 | 30.2 |
| Treatment plan | 95.3 | 46.5 | 30.2 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
