Using retrospective data of the literature, a power analysis (Power and Precision software, Biostat, Englewood, NJ, USA) was conducted that indicated that detection of differences between 2D radiographs and 3D CBCT images could be obtained with at least 35 defects at a power of 0.8 (alpha = 0.05). Thus, this study was conducted using 12 dry skulls with maxilla and mandible and 35 artificial defects (dehiscence, tunnel [furcation defect level III], and fenestration) which were created on incisors, premolars and molar teeth separately using burs.
The skulls were obtained from different museums in our country. All skulls were dated back 10th Century from different parts of country which were approved to be used for scientific study that were given by City Culture and Tourism Authorities which are connected to Anadolu Civilization Museum.
In total 14 dehiscences, 13 fenestrations, eight tunnel and 16 without periodontal defect were used in the study. These were randomly created on dry skulls. For soft tissue simulation, maxilla and mandible were covered by double layers of boxing wax (Fig. 1). The defects were created by periodontal consultant (NB) in line with Mengel et al’s study [21]. The consultant noted the periodontal defects and these were used as the Gold standard for radiographic evaluation. The periodontal defects were created using high-speed equipment with copious air/water spray and rounded diamond burs (KG Sorensen, Zenith Dental ApS, Agerskov, Denmark).
Dehiscences
Deshiscences were prepared in 5 molars, 4 premolars and 5 anterior teeth. The buccal bone in the coronal region of the teeth was removed until parallel walls until the walls are paralleled. The dehiscences had a standard dimension, approximately 10 mm height and 3 mm width from enamel-cement junction of the teeth (Fig. 2) 14.
Fenestrations
Fenestrations were prepared in 5 molars, 4 premolars, 4 anterior teeth both in maxilla and mandible. The buccal bone in the central thirds of the tooth was removed until the walls ere parallel. The fenestrations had a standard dimension, approximately 4 mm height and 3 mm width (Fig. 2).
Tunnels
All tunnel defects were prepared in mandibular molar teeth. The buccal bone lingual bone in the furcation region was removed until a continuous defect was produced. The lowest point of of the furcation was prepared as diameter of the bur, approximately 2 mm height from the furcation roof (Fig. 2).
Radiographic imaging
Each skull were exposed using a Planmeca Promax CBCT (Planmeca, Promax 3D max, Helsinki, Finland) and a Digora photostimulable phosphor plates (PSP). CBCT exposures were made in 96 kVp and 12 mA at 0.100-mm3 voxel size. The field of view was 5 cm in diameter and 5, 5 cm in height. Slice were 1024x1024 pixels. Axial, sagittal, cross-sectional images were reconstructed for all skulls, and 3D reconstructions were used as necessary (Fig. 3).
In addition to the CBCT images, a set of digital intraoral standardized periapical images was obtained. The radiographs were obtained with an intra-oral X-ray system operating at 70 kVp, 8 mA by Evolution x3000-2c (Grugliasco, Italy) and a phosphor plate digital system (Digora Soredex, Soredex Medical Systems, Helsinki, Finland). Exposure time was 0.1 s. These were taken using parallel technique with a XCP system (Rinn Co., IL, USA) device with a 12 in. cone attached. Standardization was achieved with bite blocks that were used in all radiographic examinations. The use of the paralleling technique, complemented with a positioning holder and bite blocks, minimized image enlargement and geometric distortion of the radiographs (Fig. 4).
Image evaluation
All digital intraoral images were saved in noncompressed file format (tagged image file format, TIFF). All images were displayed and evaluated on a 21.3-inch flat-panel color-active matrix thin-film transistor (TFT) medical display (NEC MultiSync MD215MG, Munchen, Germany) with a resolution of 2048 × 2560 at 75 Hz and 0.17-mm dot pitch operated at 11.9 bits. Digital intraoral images were displayed using the dedicated software of Digora imaging system (Soredex Medical Systems, Helsinki, Finland) whereas CBCT images were evaluated with its own software (Romexis 3.2, Planmeca, Helsinki, Finland). Observation conditions were optimized through use of the same computer monitor when the images were displayed. Viewing distance was kept constant to about 50 cm for the observer, and the lights were subdued during examinations.
Two dental radiologists (MEK, SK), all with 3–5 years’ experience of working with the CBCT technique examined the PSP, and CBCT images for the presence of periodontal defects in different sessions. The scores assigned by the observers were recorded by a researcher (KO) who knew the study design and had previously enhanced the images. The observers were aware that some teeth have no periodontal defects. All of the observers had access to the two views simultaneously for the intraoral and CBCT techniques. The time allocated for the observations was not restricted. Adjustment of contrast and brightness could be done, if considered necessary, using the inbuilt image display tools.
The observers were asked to define the type of the defects and also define the teeth without periodontal defects. In line with Braun’s study [10], the defects were classified being present or absent or may have been uncertain while making the diagnosis (correct, false, or questionable). In addition, all of the images were evaluated by the same examiners. For this reason, the results, positive correct and negative-correct, were summarized as “correct.” The answers: positive-false, positive-questionable, and negative-false and negative-questionable were considered “incorrect.” The level of significance was accepted at p <0.05.
All observers inter and intra evaluations were compared according to Gold standard which were created and noted by the periodontal consultant. Specificity and sensitivity for each radiographic technique were calculated. Kappa statistics was used for assessing the agreement between observers using the NCSS 2007 statistical software (NCSS and GESS, NCSS, LLC. Kaysville, UT, USA). Kappa statistics were used to determine inter and intra-observer agreement. The kappa values were interpreted according to guidelines of Landis and Koch adapted by Altman [22]. k ≤0.20 Poor, 0.21-0.40 Fair, 0.41-0.60 Moderate, 0.61-0.80 Good, 0.81-1.00 Very good. The determination of the significance level was done using the McNemar test using paired samples. Results were considered significant at p < 0.05.