This study aimed to evaluate the efficacy of fractal analysis of hand-wrist radiography in the decision of conventional or surgery-assisted rapid palatal expansion (RPE).
The study included 48 patients who underwent the RPE procedure. Study groups were as follows: group 1 (successful conventional RPE [n = 24, 5 male and 19 female patients; mean age ± standard deviation, 15.85 ± 0.97 years]) and group 2 (failed conventional RPE [n = 24, 5 male and 19 female patients; mean age ± standard deviation, 15.96 ± 1.08 years]). Fractal dimension (FD) analysis was conducted on hand-wrist radiographs of the patients for 4 different regions: the epiphysis-diaphysis line of the radius bone and the proximal, medial (MP3), and distal (DP3) phalanxes of the middle finger. A Student t test was performed to compare fractal values between the groups. A receiver operating characteristic analysis was applied to determine the optimal cutoff value of FDs. In addition, a Pearson correlation coefficient was calculated to evaluate the relationship between the fractal values and either age or hand-wrist stage in a second sample group (n = 90; age range, 8.7-18.7 years).
Fractal values of the radius, MP3, and DP3 were significantly increased in the failed conventional RPE group ( P <0.05). The optimal cutoff value of the FD for predicting the success of conventional RPE was 1.16 in the radius, 1.18 in proximal phalanxes, 1.29 in MP3, and 1.08 in DP3. There was a positive correlation between fractal values of the radius and age or hand-wrist stages ( P <0.05).
Within the limits of this study, results revealed that fractal analysis of hand-wrist radiographs might be considered a significant tool in the prediction of RPE success.
Fractal analysis of hand-wrist radiographs might help predict the success of rapid palatal expansion.
Fractal values of the radius and the medial and distal phalanxes of the middle finger were significantly increased in the failed conventional rapid palatal expansion group.
Fractal values of the radius and age or hand-wrist stages were positively correlated.
Posterior crossbite is one of the most common orthodontic problems, which can occur as skeletal, dental, or a combination of both anomalies. Diagnosis may require multiple diagnostic tools such as clinical examination, dental cast analysis, occlusograms, posteroanterior cephalograms, and cone-beam computed tomography (CBCT) imaging. The treatment varies owing to the origin of the problem, which may include dental expansion with the only dental correction or rapid palatal expansion (RPE), which targets skeletal changes. , The highest success in the RPE method is obtained in patients in the prepubertal or pubertal period when skeletal maturation is not yet complete. In the postpubertal period, the success rate of RPE decreases because of the maturation in the bone regions that are resistant to the expansion in the midpalatal suture. Inadequate expansion with RPE or failure of RPE requires an additional surgical intervention called surgery-assisted RPE (SARPE).
In the prepubertal period, RPE provides efficient results with a limited adverse effect, and in adult patients, SARPE may be the first choice until sufficient evidence exists in the literature that conventional RPE or bone-anchored RPE are reasonable options. However, the decision regarding whether to use RPE or SARPE treatment in a patient in the postpubertal period is challenging owing to the lack of a distinctive diagnostic criterion for these 2 techniques. , If the sutural fusion is mostly complete, conventional RPE would not be enough to provide adequate sutural opening, and adverse effects such as pain, necrosis of palatal mucosa, overtipping of the buccal teeth, and dehiscence or fenestration defects would likely occur. Furthermore, the next step is usually to apply SARPE, which was avoided in the first step. By contrast, implementation of SARPE as a first treatment choice in a patient whose sutural fusion is not complete will be an overtreatment because this treatment procedure is invasive, costly, and includes surgical risks. ,
Although there are specific criteria such as age, hand-wrist x-ray evaluation or CBCT to determine which technique will yield better results, each of these criteria has its deficiencies. , , The degree of maturation varies among patients and is not always correlated with the age of the patient. , , The determination of the maturation stage in the hand-wrist x-ray may provide more predictable results than chronological age in the prepubertal or pubertal period. Revelo and Fishman enounced consideration of skeletal maturity indicator in the decision of RPE and suggested the implementation of RPE in patients who have not reached the skeletal maturity indicator stage 9. However, it is not always applicable for patients in the late maturation period. CBCT, by contrast, is a useful technique in examining bone, however, it is not a routine procedure in orthodontics. Requesting CBCT in patients with RPE in routine is not preferred because of the drawbacks of the method, such as additional radiation exposure, additional cost, and subjectivity of the examiner. , Nevertheless, if a previous CBCT is available for the patient, it could provide additional information regarding the shape and interdigitation of the midpalatal suture. Therefore, CBCT may not be the first-choice technique as a predictor for the success of RPE. Other than these methods, ultrasonography was also used in evaluating midpalatal suture maturation. However, ultrasonography was not used as a prediction method but rather as a verification method used to observe the suture after the SARPE procedure.
Fractal analysis is the assessment of the fractal characteristics and the dimension of given data by simplifying the complex structure of an image and converting the complexity into a value. , Bone trabeculae also have fractal features with irregular bone plate connectivity and self-similar structure. Previous studies have indicated the efficacy of fractal analysis in evaluating the pathologic alterations in various bones such as alveolar bone or phalanxes. In addition, an association between bone mineral density and the fractal dimension (FD) was suggested. In monitoring any disease affecting bone mineral density or bone levels, such as osteoporosis, osteoarthritis, or periodontitis, fractal analysis might be used as a diagnostic tool in the quantitative determination of bone density. , , The lower the mineralization degree or bone density is, the lower the FD values. , Therefore, fractal analysis may be useful in quantitatively determining bone structure in the epiphysis-diaphysis line, and in postpubertal patients, it may provide more precise information about the patient’s general skeletal maturation, which affects the success rate of the conventional RPE.
Therefore, the hypothesis of the study was established: the examination of the hand-wrist radiograph via fractal analysis would provide insight in predicting the clinical success of RPE in patients who have passed the peak period of growth and development. To test this hypothesis, this study aimed to evaluate the efficacy of fractal analysis of hand-wrist radiographs in the decision to perform conventional or SARPE. At the same time, we will statistically evaluate whether there is any correlation between the chronological age of the patient, the hand-wrist maturation stage and the FD. To our knowledge, this is the first study to evaluate the fractal size in hand-wrist radiographs for the decision of conventional or SARPE.
Material and methods
The study protocol was approved by the medical ethics committee of Gaziosmanpasa University, and written consent forms including the use of patients’ records in scientific studies, which were taken for each patient routinely at the beginning of treatment, were checked.
The study was conducted at the orthodontics departments of Gaziosmanpasa University, Faculty of Dentistry and Bolu Abant Izzet Baysal University, Faculty of Dentistry. Records of 3000 systemically healthy patients who underwent orthodontic treatment between the years 2013 and 2018 in these clinics were evaluated. Patients with any diseases affecting their skeletal development, patients with any systemic diseases, and patients with cleft lip and palate were excluded from the study. All included patients had acrylic-bonded hyrax expanders with occlusal-coverage. In 26 of the patients, conventional RPE procedures had failed. Skeletal maturation stages were evaluated according to the method described by Björk on hand-wrist radiographs, which were taken on a Cranex 3D (Soredex Oy, Tuusula, Finland) at the beginning of orthodontic treatment. Evaluations were done by an orthodontist (S.A.) with 7 years’ experience. In total, 24 patients were in the ninth hand-wrist maturation period, and 2 were in the seventh and eighth periods. To create a homogeneous group, 24 patients in the ninth period were included in the study to form the failed-RPE (F-RPE) group. Patients in the F-RPE group had severe pain or palatal mucosa ulceration after activation of the expanders. In addition, the occlusal radiographs of these patients taken after activation revealed that there was only a small rupture in the anterior part of the suture or no opening in the entire suture.
A second age-, sex-, and hand-wrist maturation stage–matched study group, consisting of 24 patients, was created from the patients who had successful conventional RPE to form the successful-RPE (S-RPE) group. The S-RPE group consisted of RPE patients with at least 30 quarter-turns screw activation without any complications such as severe pain or ulceration and confirmed sutural opening by occlusal radiographs.
The main study groups were created as follows:
Group 1: S-RPE group (n = 24), 5 male and 19 female patients, mean age ± standard deviation (SD), 15.85 ± 0.97. All subjects were at the ninth hand-wrist maturation stage.
Group 2: F-RPE group (n = 24), 5 male and 19 female patients, mean age ± SD, 15.96 ± 1.08. All subjects were at the ninth hand-wrist maturation stage.
To evaluate the correlation between fractal values and age or hand-wrist maturation stages, a second study group was also created from the patients who were treated in the orthodontics department of Gaziosmanpasa University between the years 2013 and 2015 and who had hand-wrist radiography available in the archives. This sample group consisted of 90 patients (45 male and 45 female patients) with a mean age ± SD of 12.68 ± 2.57 years and an age range of 8.73-18.69 years. The distribution of the subjects regarding age and the hand-wrist stage is shown in Table I .
|Sex||Hand-wrist maturation stage ∗ (mean age ± SD)||Total|
|1 (9.44 ± 0.71)||2 (10.48 ± 0.79)||3 (11.22 ± 1.13)||4 (11.49 ± 0.74)||5 (12.56 ± 1.1)||6 (13.62 ± 1.84)||7 (14.32 ± 1.43)||8 (14.6 ± 1.27)||9 (17.11 ± 1.58)|
The x-ray resolution of the radiographs used in this study was 2085 × 2232 pixels with 300 dpi. FD analysis was performed on the radiographs by a radiologist (S.B.) blinded to RPE success, age, sex, and hand-wrist stage, using the Image J version 1.3 software (National Institutes of Health, Bethesda, Md). Four regions of interests (ROIs) were considered to be examined: ROI 1, epiphysis-diaphysis line of the radius bone; ROI 2, epiphysis-diaphysis line of the proximal phalanx of the middle finger (PP3); ROI 3, epiphysis-diaphysis line of the medial phalanx of the middle finger (MP3); and ROI 4, epiphysis-diaphysis line of the distal phalanx of the middle finger (DP3).
Rectangular ROIs were selected to include the entire suture as much as possible, just touching the lateral cortical bone, and including as little trabecular bone tissue as possible , ( Fig 1 ). The selected ROIs were cropped and duplicated. Thereafter, the ROI was processed with the Gaussian Blur (σ = 35 pixels) filter to blur the high- and medium-bright areas of the image, which was due to the variable thickness of the superficial soft tissue over the bone. The major differences in the density of the region concerned were thus revealed. The blurred image was subtracted from the original image, adding 128 gray tones for each pixel. Areas of different brightness in the image, whose mean value was 128 gray scale, helped to distinguish between bone marrow and trabecular structure. To reduce the noise in the image, it was first eroded with “Erode” option, and then “Dilate” option was used to expand existing areas and make them more visible. With the “Invert” option, the white areas were blacked out, and the black areas were converted to white to outline the trabecular bone. With the “Skeletonize” option, the trabecular structure was outlined as skeletal lines and prepared for fractal analysis. The “Analyze” option calculated the fractal size for trabecula outlines. The fractal size was calculated according to the box-counting method recommended by White and Rudolph because it is a simple and widely used method. , With the “Fractal box count” option, image sizes were divided into squares of 2, 3, 4, 6, 8, 12, 16, 32, and 64 pixels. The squares with trabeculae for each different sized pixel and the total number of frames in the image were calculated. These values were plotted on a logarithmic scale. The lines in the graph best fitted the point. The slope of the line gave the FD value that shows the complexity of the structure. After the measurements were completed, the FD values were recorded.
To assess the intraexaminer correlation, fractal values were assessed by re-evaluating randomly selected 20 hand-wrist radiographs again 1 month later, and the intraclass correlation coefficients showed almost perfect agreement with values >0.89 (95% confidence interval, 0.81-0.94).
Statistical analyses were performed by the SPSS (version 22.0; SPSS Inc, Chicago, Ill) software program. A Student t test was performed to compare fractal values of the radius, PP3, MP3, and DP3 between the groups. Receiver operating characteristic analysis was applied to determine the optimal cutoff value of FDs. Discriminant analysis was used to determine which variables were the best predictors of RPE success. Thereafter, the reliability of the derived discriminant function was assessed among study subjects. Pearson correlation coefficient was calculated to evaluate the relationship between the fractal values and either age or hand-wrist stage. The level of statistical significance was set at P <0.05.
G∗Power software version 184.108.40.206 (Universität Düsseldorf, Düsseldorf, Germany) was used to calculate the power of this study considering the fractal size of radius bone, and it was 99.7%.
The demographic data of the main study groups were similar ( P = 0.719) ( Table II ).