Optimization of cone beam computed tomography image quality in implant dentistry

Abstract This study was conducted to optimize the cone beam computed tomography image quality in implant dentistry using both clinical and quantitative image quality evaluation with measurement of the radiation dose. A natural bone human skull phantom and an image quality phantom were used to evaluate the images produced after changing the exposure parameters (kVp and mA). A 10 × 5 cm2 field of view was selected for average adult. Five scans were taken with varying kVp (70–90 kVp) first at fixed 4 mA. After assessment of the scans and selecting the best kVp, nine scans were taken with 2–12 mA, and the kVp was fixed at the optimal value. A clinical assessment of the implant‐related anatomical landmarks was done in random order by two blinded examiners. Quantitative image quality was assessed for noise/uniformity, artifact added value, contrast‐to‐noise ratio, spatial resolution, and geometrical distortion. A dosimetry index phantom and thimble ion chamber were used to measure the absorbed dose for each scan setting. The anatomical landmarks of the maxilla had good image quality at all kVp settings. To produce good quality images, the mandibular landmarks demanded higher exposure parameters than the maxillary landmarks. The quantitative image quality values were acceptable at all selected exposure settings. Changing the exposure parameters does not necessarily produce higher image quality outcomes but does affect the radiation dose to the patient. The image quality could be optimized for implant treatment planning at lower exposure settings and dose than the default settings.

for their periodontal imaging task.
The aim of this study is to optimize the image quality of CBCT in implant dentistry comprehensively using clinical and quantitative image quality assessments to achieve the best images at the lowest possible dose. We will measure the absorbed dose and the complete set of image quality metrics (noise, uniformity, CNR, spatial resolution, artifacts, and geometric distortion) as outlined by the Sedentex Project (2012). We will also perform an observer study to assess the diagnostic value of the images for tasks related to implant dentistry.
The outcome of the study will be a recommendation for optimized technique settings specific to implant dentistry.

| CBCT machine
A dental CBCT scanner (CS 9300,Carestream Health,Inc.,Rochester,NY,USA) at the Faculty of Dentistry, University of British Columbia, was used for the scans. A 10 × 5 cm 2 field of view (FOV) was selected in this study as it is a commonly used FOV in implant planning. The default setting for this FOV for an average adult patient is 90 kVp, 4.0 mA, 6.20 s, and a voxel size of 180 μm.

| Data acquisition protocol
Three types of phantoms were used in this study for three types of assessments: clinical image quality assessment, physical image quality assessment, and dosimetry. The CBCT scans were taken at five different tube potentials: 90, 85, 80, 75, and 70 kVp. Other parameters were fixed at 4.0 mA, 6.20 s, and a voxel size of 180 μm. The scans were taken using each phantom separately following the same protocol of exposure settings. A comprehensive analysis of dosimetry, quantitative image quality, and clinical image quality was done to select scans with good image quality at the lowest possible kVp and dose. After selection of the optimal kVp, scans were taken at nine different tube currents, 2, 2.5, 3.2, 4, 6.3, 8, 10, and 12 mA at fixed kVp, voxel size, and time. All scans were evaluated for the dosimetry, quantitative image quality, and clinical image quality to select the good quality images at the lowest possible mA and dose. The sequence of the protocol to optimize the image quality is illustrated in

| Dosimetry
The absorbed dose was measured at the five different kVp settings and nine different mA settings with fixed exposure time of 6.20 s and voxel size of 180 μm, as described above. The dose index phantom was used (SEDENTEXCT DI, Leeds Test Objects Ltd., Boroughbridge, UK; Figure 2a). The phantom is composed of six stacks of polymethyl methacrylate (PMMA) plates that simulate human tissue density (1.20 ± 0.01 g/cm 3 ). The cylinder size was of head size (160mm diameter and 176-mm height). The dose measurement method was done according to the SEDENTEXCT project (Sedentex Project, 2012). Five different regions along the phantom diameter were selected to measure the dose. The FOV was at the level of the center slice of the DI phantom. A calibrated 0.6-cm 3 thimble ionization chamber (10 × 6-0.6 CT, Radcal Corporation, Monrovia, USA) was placed into a hollow column at each of selected five regions at the level of the center slice of the phantom. Two measurements of the dose using the thimble ionization chamber were taken at five different regions along gradient of dose profile. The average of measurements was used to calculate the absorbed dose.

| Quantitative image quality
The image quality phantom, shown in Figure    To measure the AAV, the MGV in two rectangular ROIs adjacent to the rods was measured compared with the MGV of background. The spatial resolution was measured using both quantitative and qualitative methods. The qualitative method was done by visual inspection of two polymer line chart inserts located along the Z-direction and in the XY-plane. The spatial resolution was calculated quantitatively using the point spread function following the method described by Abouei et al. (2015). A square ROI was placed around the wire, and the resulting distribution was plotted using MATLAB software. The full width at half maximum was calculated. The point spread function was used to calculate modulation transfer function (MTF) using the fast Fourier transform. The spatial resolution was calculated using the frequency at 10% of MTF. Five images were taken of each arch at five different kVps (70,75,80,85,and 90 kVp). Two examiners evaluated the clinical image quality in a random order using a 5-point Likert scale (excellent, good, adequate, poor, and undetectable;Likert, 1932). The examiners who performed the assessment were blinded to the parameters of the scans and are an experienced oral and maxillofacial radiologist and an experienced periodontist. Randomized order of the images was generated using MATLAB software, version 2014 (Mathworks Inc.). Excellent, good, and adequate scores of detecting and tracing the anatomical landmark in three-dimensional assessment were considered as high image quality. Each examiner did the examination separately using the same monitor and under the same lighting conditions. The assessment of different kVp was done twice by each examiner, and they were given the choice of changing the brightness and contrast of the images.

| Subjective image quality
The images at different kVp were evaluated for the best clinical and quantitative image quality at lowest possible dose. Then, nine images of each arch were taken at nine different mA settings: 2, 2.5, 3.2, 4, 5, 6.3, 8, 10, and 12 mA. The images were assessed for the clinical and quantitative image quality following the same protocol used to assess images at different kVp.    Table 2 and at different mA settings in Table 3.

| Subjective image quality assessment
The landmarks that had a score of 3 or more were considered to have  (Figures 14 and 15).
The optimal mA setting was 3.2 mA where all maxillary and mandibular landmarks had good quality.

| Optimization
For kVp optimization, the image quality metrics showed more improvements at higher kVp (reduced noise and artifacts, slightly better spatial resolution), along with increased dose. The CNR for high contrast materials was best at 85 kVp in the image quality phantom, which was mirrored by the skull phantom results indicating that only 85 kVp and higher could produce diagnostic images for the mandible.
From these combined findings, 85 kVp was selected as the optimal value.
For mA optimization, image quality metrics also showed improvements for higher mA for noise and CNR, but improvements for artifacts and spatial resolution, the lower mA values were best. For the skull phantom, all landmarks were visible for 3.2 mA and higher.
Selecting 3.2 mA would result in slight improvements in image uniformity compared with the default setting (4 mA) and reduced dose, with slightly more image noise; other parameters such as spatial resolution, geometric distortion, and artifacts would be unchanged, and the loss

| DISCUSSION
The measurement of the dose with changes in kVp and mA was an important approach to control the applicability of the results as the      In implant dentistry, treatment planning requires assessment of bone volume, quality, orientation, as well as the local anatomy (Bornstein, Scarfe, Vaughn, & Jacobs, 2014). In our study, a natural bone human skull was used, and almost all anatomical landmarks required for implant treatment planning were present. Tracing the inferior alveolar nerve canal is an essential task for implant treatment planning but was not possible to do in the skull we used; the canal was visible only at some cross sections but could not be traced continuously. Miles, Parks, Eckert, and Blanchard (2016) studied the visibility of mandibular canal in CBCT scans, and it was visible in only 56% of the studied subjects, with younger patients (47-56 years) and females showing lower visibility than older patients (65+ years) and males.
When the visibility was further studied according to the location, at the premolar site, older males had higher visibility, whereas females had less visibility, compared with molar sites (Miles et al., 2016).
Presurgical assessment of the buccal plate thickness is needed to plan the surgical technique in cases of ridge preservation or immediate implant procedures. In a study by Timock et al. (2011), the visibility of the buccal bone thickness and height was evaluated in CBCT. Measurements were verified directly on cadavers, with higher reliability on measurements of buccal bone height than thickness; this finding was consistent with our study. However, we found that the buccal plate thickness of mandibular canine was difficult to identify in our head phantom. Other anatomical landmarks in general had adequate visibility in our study even for the smaller structures such as the antroalveolar intraosseous anastomosis at lateral walls of the sinuses as well as the superior and inferior lingual foramina.
In this study, a 5-point Likert ranking was used to assess the diag- does not allow for multiple preset technique factors specific to each arch, 85 kVp and 3.2 mA were selected as the optimal settings for the clinical image quality assessment that works for both arches.
In conclusion, the image quality can be optimized at lower dose by reducing the exposure settings as compared with default settings.
This study included a comprehensive method to optimize CBCT image quality using dose measurement, quantitative image quality assessment, and clinical image assessment. The optimization of the images is affected by the dose and should be measured together to obtain adequate diagnostic value of images at lowest possible dose complying with ALARA principle. The optimization should be task specific as different tasks may require different settings to produce the required diagnostic value. The assessment of clinical as well as quantitative image quality is required to ensure that adequate diagnostic value is obtained.