The purposes of this study were to compare the asymmetry index using panoramic radiography and cone-beam computed tomography for detecting mandibular posterior asymmetry and to evaluate the diagnostic value of the asymmetry index on panoramic radiography.
A total of 43 patients were included in this study. Ten mandibular posterior distances were measured using panoramic radiography and cone-beam computed tomography, and 10 asymmetry index values were calculated. The reliability of each asymmetry index was assessed. For evaluating validity of each asymmetry index using panoramic radiography, the paired t test and the Bland-Altman analysis were used. The accuracy of the asymmetry index and the area under the curve of receiver operator characteristic were calculated.
The asymmetry index of total ramal height showed good reliability (ICC, >0.888). In condylar height 1, specificity and negative predictive value were low (0.08 and 0.17, respectively), 95% limits of agreement were ±17.9%, and area under the curve was 0.484. In total, ramal height accuracy was 0.86, and areas under the curve were 0.926 to 0.957.
For detecting asymmetry of the condyle region, the asymmetry index using panoramic radiography had little diagnostic value, and we recommend using cone-beam computed tomography images. However, the asymmetry index for total ramal height showed good reliability and relatively higher validity, and its diagnostic value was excellent.
Even in CBCT images, condylar height 1 shows low reliability.
For detecting asymmetry of the condyle region, it is recommended to use CBCT.
Asymmetry index of total ramal height showed good reliability and validity.
Craniofacial asymmetry in varying degrees is a common finding in the human skeleton. Internal derangements of the temporomandibular joints, degenerative joint disease, rheumatoid arthritis, trauma, congenital anomalies, and craniofacial syndromes can all cause facial asymmetry. For successful orthodontic treatment of these patients, accurate diagnosis of facial asymmetry is necessary.
Panoramic radiography has conventionally been used for orthodontic diagnosis. It is simpler and more economical, and has a lower radiation dosage than cone-beam computed tomography (CBCT), which has progressively been considered the gold standard imaging technique. The posterior mandible and specifically the ramal region can be measured by panoramic radiography. However, panoramic radiography is affected by magnification, image distortion, and superposition of different anatomic structures, and it has low resolution. Despite these limitations, in many studies on facial asymmetry, condylar or ramal asymmetry is determined using only panoramic radiography.
Many studies have assessed the reliability and validity of panoramic radiographs for measuring condylar and ramal asymmetry. In terms of reliability, some authors concluded that vertical measurements of the height of the condyle or ramus could be reliably assessed on panoramic images with acceptable reproducibility. However, another study showed that the condylar head and neck were the least reliable regions. To assess validity, for the first time, Habets et al suggested using an asymmetry index, comparing vertical differences between the left and right condyles and rami. By changing the position of the mandible horizontally 10 mm or less, vertical difference can be changed less than 6%. Therefore, they suggested that differences greater than 6%, which is 3% in the asymmetry index, can be considered true vertical asymmetry. Sadat-Khonsari et al also found that asymmetries of more than 6% are probably not due to patient positioning in the panoramic machine. Using the 6% cutoff as those studies suggested, the accuracy of panoramic radiography, including sensitivity and specificity for detecting asymmetry with anatomic phantoms and skulls as standards, varied from 13% to 100%.
Although many studies have reported on panoramic radiographs for measuring mandibular posterior asymmetry, it is difficult to compare findings across studies with different assessment points and levels of accuracy. As of yet, no studies have been conducted to compare panoramic radiography with CBCT, except for 1 study with a small sample because of the use of cadaver heads. Therefore, 1 aim of this study was to compare the asymmetry index using panoramic radiography and CBCT. The other aim was to assess the diagnostic value of the asymmetry index using various measurement points in panoramic radiography.
Material and methods
For sample size calculation, a pilot test was executed with 20 patients who visited the Department of Orthodontics of Dankook University Dental Hospital in Cheonan, South Korea. To achieve power of at least 80% using the paired t test with a significance level of 0.05 and using effect sizes calculated from differences (mean, 0.342; pooled standard deviation, 0.780), at least 43 patients were needed. Those who had syndromic disease (eg, cleft lip and palate) and those with a history of mandibular surgery were excluded from this study. All panoramic radiographs of low quality, unclear images due to anatomic structure superposition, and only a partial view of the condyles were also excluded. After the clinical examination was performed and the panoramic radiograph was taken for screening, CBCT images were taken of each patient by a well-trained radiologic technologist on the same day for diagnostic reasons and for adequate treatment plans when the patients had moderate-to-severe skeletal discrepancies, were planned for orthognathic surgery, or needed evaluation of alveolar housing including buccal bone height and width, and to identify circumaxillary sutures for effective maxillary protraction and midpalatal expansion. Informed consents were taken from all patients.
Of all the patients who visited between 2012 and 2014, 66 had both panoramic radiographs and CBCT images. Five patients had clefts, and 5 had a history of mandibular surgery. The quality of the panoramic radiographs of 7 patients was low at the sigmoid notch area; 3 panoramic radiographs showed only a partial view of the condyles, and the panoramic radiographs of 4 patients were taken with other panoramic units. A total of 23 patients were excluded according to our criteria, and 43 patients (23 female, 20 male; mean age, 21.8 years; minimum, 13.2 years; maximum, 49.8 years; SD, 6.7 years) were included retrospectively in this study.
All panoramic images were taken with the Proline EC (Planmeca, Helsinki, Finland) panorama unit and captured on photostimulable phosphor plates. To minimize head positioning errors, the patients bit the mouthpiece, and their heads were positioned by 2 guide arm and a triple laser guideline. Parameters included tube voltage of 60 to 80 kV, tube current of 2 to 12 mA, total filtration of 2.5 mm aluminium, and exposure time of 18 seconds. FCR Capsula XL (Fujifilm, Tokyo, Japan) was used to process the images.
The CBCT images were obtained with the patient in the sitting position using Alphard VEGA scanner (Asahi, Kyoto, Japan). Scan mode was C-mode, which had a field of view of 200 × 179 mm, exposure time of 17 seconds, voxel size of 0.39 mm, tube voltage of 60 to 110 kV, tube current of 2 to 15 mA, and total filtration of 2.8 mm aluminium.
The CBCT images were reconstructed with the “volume render” tab in Invivo 5 (version 5.2; Anatomage, San Jose, Calif). To trace the CBCT images, reorientation with 3 orthogonal planes was performed. According to Markic et al, the orientation of each slice was standardized to intersect the center of the coronoid process, the condylar process, and the gonial angle ( Fig 1 ). After reorientation was completed and anatomic structures interfering in measurement were removed, both left and right reoriented volume images that included 5-mm and 10-mm grids for precise size settings were each captured in a JPEG file and merged using Adobe Photoshop CS5 extended (version 12.0; Adobe Systems, San Jose, Calif) ( Fig 2 ).
Acquired panoramic and CBCT images were traced and measured by 1 examiner (Y.S.L.) using V-ceph (Osstem, Seoul, Korea). Before the prepared CBCT images were traced, image size was set by the included grid. To measure the mandibular posterior region, 2 lines and 10 points were defined. Lines 1 and 2 were tangential lines, each along the posterior and lower border of the mandible. Ten points were defined as follows: condylion, the most superior point on the condyle (Cd); sigmoid notch, the deepest point of the sigmoid notch (Sg); gonion 1, the point on the bony contour of the mandibular angle determined by bisecting the angle between lines 1 and 2 (Go1); gonion 2, the intersection between lines 1 and 2 (Go2); inferior gonion, the most inferior point of the mandibular angle (Inf.go); points X and Y, the most lateral points of the ramus (X amd Y); and Cdʹ, Sgʹ, and Go1ʹ (points transferred to line 1) ( Fig 3 ). On both the left and right sides, 10 vertical variables (condylar heights 1 and 2, ramal heights 1 and 2, and total ramal heights 1-6) were measured ( Fig 4 ).
Finally, for the condylar, ramal, and total ramal heights, the asymmetry index in the panoramic radiographs and the CBCT images were calculated using the following formula described by Habets et al.
Statistical calculation was performed using SPSS statistical software (version 21; IBM, Armonk, NY) and MedCalc software (version 15.8, 64 bit; Mariakerke, Belgium). After at least 4 weeks, the asymmetry index was recalculated, and the intraclass correlation coefficient (ICC; 2-way random single measures) was used to assess the reliability of the asymmetry index on the same panoramic radiographs and CBCT images. Using Dahlberg’s formula, √∑d2/2n
∑ d 2 / 2 n
, where d is the difference between 2 measurements and n is the number of double determinations, random error was calculated. An agreement analysis for CBCT within itself was conducted for measurement error.
To check the normality of the asymmetry index, the Shapiro-Wilk test was used. According to normality, statistically significant differences between the asymmetry indexes on the panoramic radiographs and CBCT were evaluated using a paired t test or the Wilcoxon signed rank test. To assess agreement between CBCT and panoramic radiography, the Bland-Altman plot analysis was used, and 95% limits of agreement and mean differences were calculated. Using the asymmetry index on the CBCT as the gold standard, left-to-right differences more than 6%, or 3% of the asymmetry index value, were regarded as true asymmetry. Accuracy of the asymmetry index on the panoramic radiography was analyzed by 2 × 2 tables. The sensitivity, specificity, accuracy, positive and negative predictive values, and positive and negative likelihood ratios were calculated. Receiver operating characteristic (ROC) curves and area under the curves (AUC) of the asymmetry index on the panoramic radiography were used to evaluate its diagnostic value.
Table I shows the ICC values, the Dahlberg values for the asymmetry index on the panoramic radiography and CBCT, and agreement between CBCT measurements. When CBCT was used, reliability of asymmetry indexes was good (ICC, >0.853). However, ranges of the 95% limits of agreement in condylar height 1 and ramal height 1 showed relatively greater values than the rest of the measurements. For all vertical variables, CBCT was more reliable than panoramic radiography. When panoramic radiographs were used, the ICC and Dahlberg values of condylar height 1 and ramal height 1 showed much lower reliability values than did condylar height 2 and ramal height 2, which showed good reliability ( Table I ). All total ramal heights on the panoramic radiography showed good reliability (ICC, >0.888; Dahlberg, <0.89).
|Asymmetry index||CBCT||Panoramic radiography|
|ICC||Dahlberg||Mean difference (%)||Range of 95% limits of agreement (%)||ICC||Dahlberg|
|Condylar height 1||0.914||3.02||−0.9||−9.2-7.4||0.767||4.24|
|Condylar height 2||0.981||0.66||0||−1.8-1.9||0.947||1.08|
|Ramal height 1||0.853||1.07||0.1||−2.9-3.1||0.736||1.37|
|Ramal height 2||0.976||0.32||0.04||−0.86-0.94||0.945||0.48|
|Total ramal height 1||0.911||0.76||0||−2.2-2.1||0.888||0.89|
|Total ramal height 2||0.989||0.26||0.04||−0.67-0.76||0.975||0.38|
|Total ramal height 3||0.99||0.25||0.07||−0.62-0.77||0.973||0.4|
|Total ramal height 4||0.99||0.25||0.05||−0.64-0.73||0.974||0.38|
|Total ramal height 5||0.989||0.26||0.06||−0.67-0.79||0.974||0.38|
|Total ramal height 6||0.988||0.28||−0.04||−0.82-0.74||0.968||0.42|