Determination of spatial resolution
Horizontal resolution phantoms of staircase shape were used to obtain images of horizontal lead line pairs in the region corresponding to a width of 48 mm at 4-mm intervals. They were located at the incisor, premolar, molar, and the temporomandibular joint (TMJ) areas. One phantom comprised 13 steps, each containing a pair of horizontal lead line pairs of the same value. Four horizontal resolution phantoms were used each for the lead line pairs of 1.88, 2.32, 2.58, and 3.19 lp/mm. These values were determined based on the IEC Standard 4, which uses lead line pairs of 1.6–3.0 lp/mm for evaluating panoramic radiography [18]. Nuclear Associates model 07-501 SER. NO.12913® (Fluke Co., Cleveland, OH, USA) was used, laser cut and placed at each step.
Four vertical resolution phantoms with vertical lead line pairs placed at an oblique angle to the same width of 48 mm were also designed, each for the lead line pairs of 1.88, 2.32, 2.58, and 3.19 lp/mm. Ligature wires were used at the boundary to mark the 4-mm interval horizontally.
To locate the prepared phantoms, an arch shaped phantom stand was used. It was designed to have square-shaped holes in four positions, incisor, premolar, molar, and TMJ. The data of the incisor, premolar (central points of the cusps of mandibular first premolar and mandibular second premolar), molar (central points of the mesiobuccal cusp of mandibular first molar and the distobuccal cusp of mandibular second molar), and TMJ were determined by previously developed ball type panorama phantom [14]. A space of 20 mm was left under the phantoms to overcome the limitation of overlapping images of the lead line pairs at the lowest step due to the phantom stand being placed on the opposite side of the actual image of the phantom itself to reduce the ghost image.
Panoramic radiographs of the phantoms were obtained using OP-100® (Instrumentarium Dental, Tuusula, Finland), PCH-2500® (Vatech, Gyeonggi, Korea), and Rayscan α-P® (Ray, Gyeonggi, Korea) at Seoul National University Dental Hospital, Seoul, South Korea. The images were obtained with optimal parameters according to the user manual for imaging adult males, which are regularly used in the department. The imaging parameters were 73 kVp, 10 mA, and 17.6 s for OP-100®; 73 kVp, 10 mA, and 13.5 s for PCH-2500®; and 73 kVp, 10 mA, and 14.0 s for Rayscan α-P®. The midpoint of the centerline of the phantom stand was arranged at the center of the incisive notch. A tripod water level was used to position these accurately. Attenuation by the skull was reproduced with an 0.8-mm copper plate on the X-ray source, according to the recommendations of IEC Standard 4 [18]. According to the standard, a 6-mm aluminum plate should be attached to the front of the phantom to mimic attenuation by soft tissue. However, the entire phantom could not be covered uniformly due to structural limitations such that it overlapped with the copper plate and was placed in the X-ray source. Images of the phantoms were obtained three times in the same position to compensate for errors in image acquisitions.
The images of resolution phantoms were evaluated twice by the consensus of two oral–maxillofacial radiologists with over 20 years of experience. Following the random arrangement of the images, the readable range numbers (1 ~ 13, numbered from the lowest area) were determined. Images were rearranged randomly after 2 weeks and reevaluated to minimize error. The images of one phantom resolution image were acquired three times, and each image was read twice, resulting in six results for each resolution phantom for one region. The value that could be read more than four times was determined as the readable line pair value for that area.
Determination of Ball Distortion Rate
The ball-type panorama phantom proposed in a previous study was used [14]. This phantom was designed to obtain 704 metal ball images without overlapping by acquiring a single image. The balls were closely positioned in a 46-mm wide jawbone shape.
The same panoramic radiography systems (OP-100®, PCH-2500®, and Rayscan α-P®) and parameters were used. The midpoint of the centerline of the ball-type phantom was positioned at the center of the incisive notch. A tripod water level was also used. For the ball-type panorama phantom, additional bone and soft tissue attenuation was not reproduced to compensate for the X-ray attenuation caused by acrylic resin placed on the ball. The exposure was repeated three times to minimize errors.
A program was created using MATLAB® (Mathworks Inc., Natick, MM, USA) to analyze the obtained ball-type panoramic phantom images. To clearly emphasize the boundaries of the obtained images of the balls, a “noise reduction filter” was applied and then “ellipse detection” was applied with a “universal threshold” to detect the boundaries of the balls obtained in circular or elliptical shapes on the images. The lengths of the long and short axes of the balls were measured using the boundaries. The differences in ball distortion ratio of the two lengths was used as an index representing the degree of ball deformation (ball distortion rate).
By applying various distortion rate thresholds from 5–50% at 5% increments, the lowest satisfied ball distortion rates were determined as the final-ball distortion rate.
Data analysis
The theory behind the rationale of the study was that the resolution value at the same site evaluated by each of the horizontal and vertical line pair phantoms would show different results. However, the ball distortion rate of the part that satisfies clinically meaningful criteria (evaluated by each spatial resolution measurement method) will not show a significant difference in its mean value. Once this theory is validated, the reference ball distortion factor, which can satisfy both the vertical and horizontal diagnostic values, can be presented as a criterion for the focal layer.
To analyze the ball distortion rate corresponding to the area showing clinically meaningful quality, the reference line pair values with clinically desirable qualities from the previous study were used [17]. The reference line pair values were 3.19 lp/mm in the incisor, 2.32 lp/mm in the premolar and TMJ, and 1.88 lp/mm in the molar region.
The ball distortion rates for the same site that satisfied the reference line pair values using a horizontal or vertical resolution phantom were selected and divided into the horizontal phantom and vertical phantom groups.
Statistical analysis was performed using SPSS 21.0® (IBM Corp., Armonk, NY, USA). Values for all parameters were expressed as numbers. Levene’s test was used to determine whether data were normally distributed, and Student’s paired t-test was used to compare the ball distortion rates satisfying each line pair value. A P-value of < 0.05 was considered significant.