Accuracy of cone beam computed tomography in predicting the linear dimensions of impacted maxillary canines- A clinical study

INTRODUCTION

Anthropoidal craniofacial patterns were first analyzed by recording different dimensions of dry skulls. The first measurements were based on osteologic landmarks. Later, measurements were made directly on patients by palpation, and, with the invention of the x-ray machine, measurements were made on cephalometric radiographs. These measurements are used in orthodontics to diagnose the malocclusion and also help in planning the appropriate treatment. One major limitation with conventional radiographic techniques is that information can be lost when a 3-dimensional structure is compressed into a 2-dimensional image andcompromise diagnostic accuracy.²

CBCT has been developed for 3D imaging of the craniofacial field. It provides low radiation, rapid image scanning with radiographic and 3-D volumetric data for each patient.¹⁷ The computer software addresses the projection effect, resulting in undistorted 1:1 measurements. This contrasts with traditional imaging, which has some projection error because the anatomic regions are at varying distances from the film.

Studies carried out to compare CBCT with other methods of measurement have shown minimum distortion and errors.^18-21 These studies were carried out on animal heads,¹⁸ extracted teeth,¹⁹ dry human skulls²⁰ and embalmed cadavers.²¹ Although the accuracy and reliability of linear dental measurements on CBCT have been well assessed, till now, no studies have been carried out to assess the accuracy and reliability of CBCT in predicting the size of unfavorably impacted teeth in human subjects.

Afterthird molars, the maxillary canine is the most frequently impacted tooth. Ferguson found the incidence in orthodontic patients to be as high as 23.5%.²² Palatal impactions are more prevalent (85%) than labial impactions (15%).²³ The early detection of impactions is critical to the success of orthodontic treatment. Ericson and Kurol found that the more mesially located the crown of an impacted canine, the lower the likelihood of its eruption after deciduous extraction.²⁴ Lindauer et al, developed a method for predicting eruption after deciduous extraction.²⁵ According to their classification, if the canine was radiographically located in sector III or IV there was a high percentage chance that it will be unfavourably impacted.

Root resorption is not only the most common sequelae of canine impaction but the most difficult to treat. Although the maxillary lateral incisor root is the most commonly affected by impaction of the canine,¹there is also evidence that impacted canines can cause root resorption of the central incisors.^26,27

The purpose of this study was to compare the linear measurements of impacted maxillary canines made on CBCT with the linear measurements made with digital calipers. The CBCT scans taken for the purpose of this study could also be used to exactly localize the impacted canine, assess the proximity of the canine to central and lateral incisor roots and help to prevent the unwanted sequelae of canine impaction.

Material and Method

This clinical study was conducted in the Department of Orthodontics and DentofacialOrthopedics, Government Dental College and Research Institute, Bengaluru. ¹⁸ patients visiting the Department of Orthodontics, who fulfilled the selection criteriawere included in this study. Informed consent was obtained from the parent/guardian if the patient was a minor. The study was approved by the institutional ethical committee and review board, GDCRI, Bangalore. The individuals were selected based on the following criteriaInclusion Criteria:

• Age group of 13 years and above with complete root formation of impacted maxillary canines
• Maxillary impacted canine classified as sector III or IV²⁵
• Patient willing for surgical extraction of impacted maxillary canines.

Exclusion criteria: 

• History of previous orthodontic treatment.
• Presence of any syndrome. 
• Presence of any systemic disease.
• History of trauma or any other contraindication for radiation exposure or surgery.

A preliminary examination of the patients was carried out and out of eighteen patients selected, sixteen patients had unilateral unfavorably impacted maxillary canine²⁵ and two had bilateral unfavorably impacted maxillary canines²⁵ (Total 20 teeth).

CBCT scan was taken of the quadrant of interest only using KODAK 9000 3D Extraoral Imaging System (Carestream Health, Rochester, New York) (Fig.3,4) and the data was exported as DICOM images to 3D Imaging Software (Carestream Health, Rochester) (Fig.5). After analysis, atraumatic surgical removal of impacted canines was done (Fig.6,7). Two teeth were excluded from the sample as they were sectioned with a bur to facilitate removal.

All linear measurements (mesiodistal, labiolingual and total length of tooth) were carried out on the DICOM images. The field of view (FOV) was confined to the portion of the anterior maxillary arch to visualize the entire unfavorably impacted canine. This enabled us to have a voxel size of 98 µm for greater accuracy of measurements. Oblique slicing was used to carry out the desired measurements on CBCT scan. First, the section was aligned along the long axis of the tooth and total tooth length and labiolingual width were measured. Then the section was aligned perpendicular to the long axis of the tooth and the mesiodistal width was measured.

The total tooth length was measured from the most incisal to the most apical part of the tooth, labiolingual width was measured from most apical point on the CEJ on both labial and lingual sides,mesiodistal width was measured from most occlusal point on the CEJ on both mesial and distal sides.

After the surgical removal of impacted canines, the same measurements were made using digital calipers (RSK Technologies, China) (Fig.10,11). The physical measurement with calipers is considered the gold standard for the purpose of this study. The teeth were collected (Fig. 12) and the same points were used for measurement with calipers as those used for the CBCT scans (Fig. 13,14). The mesiodistal width measurement was made by applying the beaks of the calipers from the apical aspect (Fig. 15).

The readings obtained from CBCT scans and digital calipers were compared using paired-t test & the accuracy of the CBCT machine & software was determined.

The data obtained were subjected to statistical analysis. Means, standard errors, and standard deviations were tabulated, with the paired t test (calculated with the help of SPSS (Statistical Package for Social Sciences Software). The level of significance was set at p ≤ 0.05. Intra and interexaminer reliability was assessed using the Karl Pearson coefficient of correlation.

Results

This study was conducted to compare the linear measurements of impacted maxillary canines measured by CBCT scans with physical measurements from digital calipers. Comparison of CBCT and Digital Caliper methods in 1st and 2nd measurements of labiolingual width, mesiodistal width and total tooth length of examiner 1 by Paired-t test (Table 3) showed no statistically significant difference for labiolingual width but showed a statistically significant difference for mesiodistal width and total tooth length in 1st measurements. In the 2nd measurement, no statistically significant difference was seen for labiolingual width and mesiodistal width but a statistically significant difference was present for total tooth length. In average no statistically significant difference was seen for labiolingual width and mesiodistal width but a statistically significant difference was seen for total tooth length.

Comparison of CBCT and Digital Caliper methods in measurements of labiolingual width, mesiodistal width and total tooth length of examiner 2 by Paired t test showed a statistically significant difference for total tooth length (Table 4).

Comparison of 1st and 2nd measurements of labiolingual width, mesiodistal width and total tooth length by examiner 1 in CBCT and Digital Caliper methods by Paired t test showed a statistically significant difference for total tooth length in the digital caliper measurements (Table 5).

Comparison of examiner 1 (average of 1 and 2 measurements) and examiner 2 of labiolingual width, mesiodistal width and total tooth length in CBCT and Digital Caliper methods by Paired t test showed a statistically significant difference for total tooth length for the CBCT measurements (Table 6). Intra and inter observer reliability was excellent as coefficient of correlation was above 0.92 for all digital caliper and CBCT measurements (Table 7-10).

Discussion

An accurate prediction of the mesiodistal diameter of the erupting permanent teeth, in particular, the unerupted canines and premolars, is essential in orthodontic diagnosis and treatment planning. A discrepancy between the space required and the space available will result in excessive spacing or crowding. An accurate estimate of available space is necessary for making correct decisions in orthodontic treatment planning.

Numerousstudies have been conducted to compare the accuracy of various methods of predicting the MDD of the unerupted permanent canines and premolars, but no single method has been shown to deliver high accuracy and reliability. Some studies tend to underestimate (Foster⁹ , Ballard¹², Nance^[10]), and others overestimate their predictions (Moyers^,8 Tanaka, Johnston¹¹).Legovic et al studied⁸ methods and found that Tanaka andJohnston mixed dentition analysis proved to be the most reliable with correlation coefficient between 0.54 and 0.77.⁴³

CBCT method eliminates population variations, since measurements are made individually rather than regression models or tables of average tooth sizes that might not be valid for certain populations. Baumgaertel et al,²⁰ Periago et al,²⁹ Ballrick et al,³⁰ Brown et al,³³ and Damstra et al³⁴ reported that CBCT linear measurements had a tendency to underestimate the reference values. But Lagravere et al found that CBCT linear and angular measurements showed no statistically significant differences between the coordinate measuring machine and CBCT, which can produce a 1-to-1 image-to-reality ratio.

Baumgaertel et al conducted a study to investigate the reliability and accuracy of dental measurements made on CBCT reconstructions²⁰ and found that both the CBCT and the caliper measurements were highly reliable.In their study, Baumgaertel et al used digital calipers calibrated to the nearest 0.01mm and imaging software calibrated to the nearest 0.1mm. The same calibrations were used in the present study.

A study by Sherrard et al carried concluded that CBCT scans are atleast as accurate and reliable as periapical radiographs for tooth and root-length determinations. A study by Ye et al¹⁹ concluded that it increasing voxel sizes during scanning, the volume measurements of teeth tended to belarger. In the study by Ye et al, laser scanner was compared with CBCT and was found to be more accurate than CBCT but significant pre- and postprocessing is required with this technique. The voxel size used in the present study was 0.098mm which according to Ye et al is better than larger voxel sizes for dental measurements.

A simulation study by Nguyen et al⁷ concluded that CBCT can be used to measure the MDD of unerupted teeth in orthodontic patients with smaller errors than any other method.Models simulating unerupted teeth cannot accurately replicate the non-homogeneous supporting tissues of teeth and hence cannot give a true picture of the complex interaction of the X-rays and the tissues. The results of the present study are more reliable as human subjects were used and hence, a true picture of tooth and bone anatomy could be obtained

For measurement purposes, at the mesiodistal portions of 2 adjacent tooth crowns, the enamel surrounding artifacts produced by the partialvolume effect overlap, and are inconspicuous; this might not have a great influence on mesiodistal width measurements. But if there is a space between 2 teeth, the enamel surrounding artifacts might have an influence on measurements with larger voxel scans. In small voxel scans, the PDL between the root and the alveolar bone can be obtained clearly; this could help for root segmentation. But in larger voxel scans, the periodontal membrane cannot be displayed clearly, so it is difficult to segment the root causing greater deviations for volume measurements. For root resorption to be assessed accurately in vivo, it is recommended to choose small voxel scans and use the same CBCT machine with the same scan parameters before and after treatment.

This investigation included CBCT images for linear quantification of total tooth length, labiolingual tooth width and mesiodistal tooth width compared with digital caliper measurements which is the gold standard for the quantification of hard tissues.

CBCT labiolingual tooth width and mesiodistal tooth width measurements were accurate. Most canine teeth had concavities at the mesial and distal CEJs, so care should be taken to measure mesoidistal tooth width from the apical aspect to avoid incorrect measurements. Total tooth length measurements derived from the CBCT images were accurate compared with digital caliper. The differences between measurements were statistically significant. Tooth lengths were systematically overestimated by an average of 0.14 mm. This might be because of the following reasons:

1. Computed tomography uses the Hounsfield unit as its unit of measure. The reconstructed tooth volume contains artifacts, which can lead to larger tooth volumes and are difficult to eliminate during the 3D reconstruction procedures.

2. Difficulty in identifying root apices when the root is separated from bone by just the PDL space.

3. Differences in orientation.

4. Imageresolution and human errors in landmark identification.

5. Patientmovements, can cause blurring of the 3-dimensional image

Although the mean difference for CBCT tooth length were statistically significant, all mean differences for CBCT tooth length were within 0.25 mm (clinically insignificant).Intra-examiner and Inter-examiner correlations were high for all tested measurements.

Although this study demonstrates a promising method of accurately measuring the MDD of unerupted teeth, it has yet to be determined whether the added precision of this method over the alternatives discussed would justify exposing a patient to radiation solely for this purpose. The American Dental Association’s Council on Scientific Affairs recommends techniques to reduce the amount of radiation received during dental radiography: the “as low as reasonably achievable” principle. Damstra et al concluded that increased voxel sizes did not result in greater accuracy of surface model linear measurements in the mandible.³⁴ Therefore, the choice of voxel size for CBCT scanning should depend on the patient’s diagnosis and treatment plan.

Conclusion

This study was done to investigate the accuracy of CBCT scan in all 3 planes of space by comparing the linear measurements of impacted maxillary canines measured by CBCT with physical measurements from digital calipers.From this study, it can be concluded that:

1. Intra and inter-observer reliability was excellent as Pearson’s coefficient of correlation was above 0.92 for both digital caliper and CBCT measurements.

2. Labiolingual and mesiodistal tooth width measurements from CBCT scans were accurate and the differences between the digital caliper and CBCT measurements were not statistically significant.

3. Total tooth length measurements derived from the CBCT images were less accurate compared with the digital calipers measurements. Tooth lengths were systematically overestimated by an average of 0.14 mm.

4. When compared with digital calipers, the CBCT scan used in the study has clinically accurate measurements and acceptable resolution at 0.098mm voxel size and hence can be used to accurately predict the diameter of unerupted teeth.

The voxel size used in the present study was 0.098 mm. In the CBCT scanner used in the present study, scans can be taken at 3 resolutions, 0.076 mm, 0.098 mm and 0.2mm.Ye et al recommend that, for high- precision volume applications (i.e., 3D computer-aided design model development, dental measurement, predicting the diameters of unerupted teeth, and evaluation of root resorption), voxels of 0.125, 0.20, and 0.25 mm are better ^[19] Further human studies with increased sample sizes may be required to gain more insight in this field.