Accuracy of measurements performed on digital panoramic radiographs with and without an extra-oral calibration object

Results: One sample t-test was used for the analyses of accuracy of measurements in vertical and horizontal direction. The most accurate vertical measurements were achieved in canine group in JPEG (6.02±0.04 mm, P=0.144) and DICOM (6.03±0.07 mm, P=0.104) formats using calibration object placed in the middle of the radiograph. The mean values of measurements in horizontal direction differed significantly from the real values (P<0.05) in all teeth groups regardless of the image format.


Introduction
Panoramic radiography is commonly used imaging modality representing a valuable diagnostic tool in the dental clinical practice [1]. Widely available, at relatively low cost, it allows visualization of dentoalveolar anatomy of both dental arches with low radiation dose [2][3][4][5]. The panoramic radiograph also depicts numerous anatomic structures outside jaws, creating further interpretation challenges for dental clinicians [1]. Additionally, pan-Cite as: Ivanović A, Mihanović F. Accuracy of measurements performed on digital panoramic radiographs with and without an extra-oral calibration object. ST-OPEN. 2020; 1: e2020.1917.11. oramic radiographs are standard examination tools for the assessment of bone architecture and are useful in planning of the initial diagnostic phase of the implant treatment [5]. It may also be indicated for the purpose of postoperative assessment and periodic review of implants [5]. Cone-beam computed tomography (CBCT) is required for preoperative cross-sectional imaging when clinical conditions indicate a need for augmentation procedures and bone reconstruction prior to the implant placement. In addition, CBCT is indicated as postoperative imaging technique only if the patient presents the implant mobility [5].  [4]. Each image format has characteristic advantages and disadvantages e.g. JPEG format can be opened with different viewers while DICOM format can be opened with specific DICOM viewer. Both image formats have to be considered while using images for clinical, educational or scientific purposes [6][7][8]. Panoramic radiography allows performance of measurements on digital images using different software but users must be aware of their characteristics (including certain imperfections).
Image distortion, due to variations in a degree of magnification in horizontal and vertical planes, may affect the accuracy of linear measurements in panoramic radiographs [9,10].
The aim of this study was to investigate the accuracy of measurements in vertical and horizontal planes/directions using an extra-oral calibration object (metal scale ruler) placed in different areas on a panoramic radiograph in two standard image formats (JPEG and DICOM). We also investigated the distortion of images in different regions of the mandible and the maxilla. We hypothesized that the extra-oral calibration of digital panoramic radiographs will improve the accuracy of measurements performed in dental practice independently of the image format.

Methods
In this test-model study the Transparent Demonstration Model (Hager&Werken GmbH, Duisburg, Germany Ref: 355641) was used. The life size modelhad 32 removable teeth which were replaced with 32 metal balls 6 mm in diameter. Each metal ball was embedded in a block of silicone impression material (ExpressTM STD, 3M ESPE Deutchland GmbH, Dental products, Neuss, Germany) and inserted in alveoli, in natural teeth position. The model with 32 inserted metal balls was imaged in a digital panoramic X-ray unit Orthopantomograph OP 200D (Instrumentarium Dental, Tuusula, Finland) in the Center for Dental Radiology X-Dent, Split, Croatia. During imaging, the model was fixed on the chin rest of the digital panoramic X-ray unit with the adhesive bandage to avoid any movement (Figure 1). It was centered in the midline of the panoramic X-ray unit and was placed horizontally perpendicular to the midline, using the laser light beam as a guide. The oil-  calibration tool, which automatically calculated the distance between the selected points to the nearest tenth of a millimeter. The same software was used for all measurements, in DICOM and JPEG image format with 100% image magnification and identical contrast and brightness. Readings were recorded by the observer and stored as computer files.

Metal ball length evaluation
All panoramic radiographs were analyzed separately by three independent observers experienced in image interpretation and all three were blind to other observers' results. One observer was general dental practitioner with 15 years of clinical experience in dentistry. The other two observers were radiographers, experienced in medical radiology and trained in the field of dental radiography. Each image set (JPEG image format without the calibration, JPEG image format with calibration object in the middle, JPEG image format with calibration object on the side, DICOM image format with calibration object in the middle and DICOM image format with calibration object on the side) was analyzed twice, in a random order, in a different sessions at 2-week interval between viewing, to eliminate memory bias. Prior to the official measurements, the observers practiced using the software on the radiographs, which were not part of this study, in order to get familiar with its functions. Each observer examined the largest horizontal and vertical distance on each of 32 metal balls, randomly, on 5 different radiographs using a mouse-driven pointer  Because of different image distortion in different panoramic radiograph regions, we divided measurements in four groups according to the teeth position in dental arch (tooth numbering): incisors (11,12,21,22,31,32,41,42), canines (13,23,33,43), premolars (14,15,24,25,34,35,44,45) and molars (16,17,18,26,27,28,36,37,38,46,47,48). presented as mean ± standard deviation and as median with range (minimum -maximum) for each group separately. One sample t-test was used for the analyses of differences between measured and real values of 32 metal balls in different teeth groups and measurement methods. The distortion ratio (DR) was calculated for measurements of each metal ball for vertical and horizontal plane as follows:

Inter-and intra-observer reliability
Overall Interclass correlation coefficient (ICC) revealed high degree of inter-and intra-reliability between the observers and measurement sessions for all results (ICC=0.997, 95% confidence interval (CI) 0.996 -0.997, P<0.001) (Figure 3). ICC between Observer 1 and Observer 2 was 0.991, between Observer 1 and Observer 3 0.992 and between Observer 2 and Observer 3 0.989, showing acceptable inter-agreements. The interpretation of the intra-observer agreement was similar to the inter-observer agreement.

Overall accuracy of vertical and horizontal measurements of 32 metal balls with or without the calibration object in JPEG and DICOM image format
The accuracy of measurements of 32 metal balls (6 mm diameter) showed differences related to the measurement method (with or without the extra-oral calibration object), group of the teeth measured (incisors, canines, premolars, molars) and the plane of the   Table 2.

Distortion ratio (DR) of 32 metal balls with or without the calibration in JPEG and DICOM image format
Image distortion is a result of specific geometric projections of panoramic radiography.
The values of the calculated metal balls DRs are in accordance with the real measurement values. Distribution of DRs with or without the calibration in different image formats are shown in Table 3.

Discussion
In this study we used the metal scale ruler placed extra-orally so that it appears in the middle of the panoramic radiograph as the most accurate calibration object [22]. In our study vertical distortion ratio varied from mean of 1.00 in incisor and canine region to 1.04 in molar region on JPEG image format with calibration achieved by the metal ruler being positioned in the middle of the panoramic radiograph, similarly as in DICOM image format. When measurements were performed in horizontal direction, it was found that differences could be related to the position of a specific tooth within each dental arch.  [19,24,27,28]. Also, the underestimation of the available bone height may lead the choosing of the smaller implant than needed, which can compromise the treatment success [24]. Therefore, in certain clinical cases it would be advantageous to use more precise radiological techniques such as new generation of dental panoramic radiographic equipment with tomosynthesis and cone beam computed tomography (CBCT) [4-6, 29, 30].
evaluated the measurement accuracy of digital panoramic radiographs using posterior mandibular implants and metal balls as reference objects and showed reliable accuracy in both modalities [14]. They did not find any significant differences in the vertical measurements of implants (the mean vertical DR=0.99) and metal balls (the mean vertical DR=0.97), respectively, and the accuracy was unrelated to the mandibular sites [14]. Many previous studies reported the accuracy of vertical measurements on digital panoramic radiographs but there is limited evidence regarding the accuracy of horizontal measurements [14,[19][20][21][22][23][24][25].
The limitation of this study was the lack of in vivo characteristics of our model. Since teeth with different positions within the dental arch appear slightly different on radiographs, mainly due to the distortion, we believe that extra-oral placing of the calibration device, so that it appears in the middle of the image, can reduce measuring errors. However, our results need to be confirmed in future in vivo studies. Peer review: Externally peer reviewed.

Funding: None.
Authorship declaration: Anita Ivanović contributed to acquisition, analysis and interpretation of the data; drafting and providing final approval of the version to be published; Frane Mihanović contributed to the design and concept; acquisition, analysis and interpretation of the data; drafting and critical revision, supervision of the study; and providing final approval of the version to be published. The authors are agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Competing interests:
The authors completed the ICMJE Unified Competing Interest form (available upon request from the corresponding author), and declare no conflicts of interest.