doi:10.1111/iej.12466

Digital radiographs displayed on different devices: effect on the detection of vertical root fractures

T. V. Vasconcelos1, G. M. Santaella1, H. A. R. Nascimento1, K. Rovaris1, G. M. B. Ambrosano2 & D. Q. Freitas1 1

Division of Oral Radiology, Department of Oral Diagnosis, State University of Campinas, Piracicaba; and 2Division of Bioestatistcs, Department of Community Dentistry, Piracicaba Dental School, State University of Campinas, Piracicaba, Brazil

Abstract Vasconcelos TV, Santaella GM, Nascimento HAR, Rovaris K, Ambrosano GMB, Freitas DQ. Digital radiographs displayed on different devices: effect on the detection of vertical root fractures. International Endodontic Journal, 49, 386–392, 2016.

Aim To evaluate whether the type of display device affects the detection of vertical root fractures (VRFs) on digital radiographs in unfilled canals and canals with fibreglass posts. Methodology Forty single-rooted human teeth were decoronated, and the root canals were prepared. The teeth were divided into 2 groups: controls (20 teeth) and with VRF (20 teeth). The VRFs were induced using an universal testing machine. Periapical radiographs of all teeth, with canal unfilled or with a fibreglass post, were obtained using the parallel technique in 3 directions (ortho-, mesio- and distoradial) on storage phosphor plates (VistaScanâ). All images were evaluated and re-evaluated after 30 days by 3 examiners on a 5-point scale using 4 different

Introduction The advent of digital radiography has made the use of desktop computers and cathode ray tube (CRT) monitors more frequent for radiological diagnosis. The

Correspondence: Helena Aguiar Ribeiro do Nascimento, Division of Oral Radiology, Department of Oral Diagnosis, Piracicaba Dental School, State University of Campinas (UNICAMP), Avenida Limeira, P.O. Box 52, Zip Code 13414903, Piracicaba, S~ ao Paulo, Brazil (Tel.: +55-19-2106-5327; e-mail: [email protected]).

386

International Endodontic Journal, 49, 386–392, 2016

devices (notebook display with full high definition resolution, desktop display with a standard resolution, 8-inch AndroidTM tablet with high definition resolution and a 9.7-inch iPadâ tablet with Retina resolution). Areas under ROC curves, sensitivity, specificity and accuracy values were compared by ANOVA. Results The weighted kappa values for intra- and interobserver reproducibility were 0.55–0.88 and 0.31–0.65, respectively. There was a significant difference (P < 0.05) in relation to the area under the ROC curve, specificity and sensitivity when unfilled canals were compared with canals with a fibreglass post; however, no difference was observed for the different devices studied. Conclusions The type of display device did not affect the detection of VRFs. Thus, the detection of VRFs can be performed using different screen sizes and resolutions. Keywords: diagnostic imaging, digital radiography, monitor, tooth fractures. Received 5 February 2015; accepted 13 May 2015

size and problems with light reflections on CRT screens led to alternative types of screens, for example liquid crystal displays (LCDs), which eliminated peripheral distortion artefacts and provided excellent spatial resolution (Balassy et al. 2005, Park et al. 2008). New types of screen technologies are now available, such as light-emitting diodes (LED), thinfilm transistors (TFT) and in-plane switching (IPS) LCDs, and they can also be used for diagnosis in dentistry. A vertical root fracture (VRF) is a longitudinally oriented fracture (American Association of Endodontists

© 2015 International Endodontic Journal. Published by John Wiley & Sons Ltd

Vasconcelos et al. Digital radiographs on different devices

2008); it is the third most common reason for extraction of teeth with endodontic treatment (Toure et al. 2011). A VRF may be misdiagnosed as localized periodontal lesion or failed root canal treatment (Patel et al. 2013), and the differential diagnosis can be difficult due to the lack of specific symptoms and radiological signs (Tsesis et al. 2010). Radiographic diagnosis is difficult and depends on the correct radiographic angulation with multiple radiographs (changing the horizontal angle) to increase the likelihood of revealing the fracture (Varshosaz et al. 2010). The few studies that have examined the effect of the display monitor on dental digital radiographs have focused mainly on the detection of carious lesions (Esmaeili et al. 2007, Ilg€ uy et al. 2009, Isidor et al. 2009). Only one study has compared the performance of different display devices for the diagnosis of VRFs; however, just 2 medical screens and 1 conventional LCD monitor were evaluated (Tofangchiha et al. 2013). Currently, several types of devices with different screen size, spatial resolution, levels of greyscale and colour depth are commercially available, and the use of mobile devices has increased; however, the effect of their use for radiographic diagnosis, particularly for VRFs, is not known. The aim of this study was to evaluate whether the display device affected detection of VRFs on digital radiographs in unfilled teeth and in teeth with a fibreglass post. The null hypothesis is that there are no significant differences on different display devices used for VRFs diagnosis.

following sequence: size 30, .05 taper, size 35, .04 taper, size 40, .04 taper and size 25, .07 taper. For further adaptation of an 1.3-mm diameter and 20mm length intracanal post, preparations were achieved with a drill at low speed (No. 3-Reforpost, Angelus, Londrina, Brazil) for two-thirds of the length of the root canal (cervical and middle). VRFs were created in 20 teeth according to the methodology used by Patel et al. (2013). Each tooth was placed in an acrylic block, and the VRF was created using an Instron universal testing machine (Canton, MA, USA) at a speed of 1 mm min 1 towards the tooth root, with a force of 500 N, which was set to stop when the root fractured. To confirm the presence of the fracture, the teeth were inspected by direct visualization (Fig. 1) and transillumination through a LED light (Ultralume 5, Ultradent Products Inc., South Jordan, UT, USA). Each root was then placed in a socket within a dry human mandible (pre-molar region), which was created with a cylindrical bur. The mandible was placed in a 2.5-mm thick acrylic container to simulate soft tissue prior to obtaining the radiographic images. Radiographs were taken in an orthoradial direction and 2 other directions by varying the horizontal angle to 15° mesial and distal, indicated by a protractor attached to the set-up. The images were obtained using a phosphor plate, size 2 (30 9 40 mm active area and 25.6 line pairs

Materials and methods Forty single-rooted human teeth (incisors, canines and mandibular pre-molars) were selected after approval by the local research ethics committee. Clinical and radiographic evaluations were performed to exclude teeth with root filling, internal or external root resorption, supernumerary roots, obliterated canals, pulpal calcifications, open apices, fractures and cracks. The crowns of all teeth were removed at the cementoenamel junction by a diamond disc cutter (Isomet 1000, Buehler Ltd, Lake Bluff, IL, USA) to eliminate identification of coronal fractures, which could be indicative of root fracture. For all teeth, the rotary NiTi MTwo endodontic system (VDW, Munich, Germany) with 350 rotations per min and 1 N was used with distilled water for irrigation. The entire length of the root canal was treated with files in the

© 2015 International Endodontic Journal. Published by John Wiley & Sons Ltd

Figure 1 Direct visualization to confirm the presence of the fracture (arrow).

International Endodontic Journal, 49, 386–392, 2016

387

Digital radiographs on different devices Vasconcelos et al.

(lp)/mm spatial resolution) with the VistaScanâ digital system (D€ urr Dental, Beitigheim-Bissingen, Germany), scanned immediately after acquisition using a VistaScanâ Perio Plus (D€ urr Dental) scanner and DBSWIN software (D€ urr Dental). All radiographic exposures were made on a GX-770 periapical machine (Gendex Dental Systems, Lake Zurich, IL, USA) at 70 kVp, 7 mA and 0.08 s and a focus-toobject distance of 40 cm. Two images were acquired for each tooth: without root filling and with a fibreglass post (Reforpost, number 2, Angelus, Londrina, Brazil) (Fig. 2). No cement or adhesive materials were used, so that the same root could be radiographed under different conditions. After acquisition, the images were exported from the software in a PNG format, using an 8-bit scale, and for the evaluation procedure, they were randomly numbered and transferred to the various displays using an external hard drive, wi-fi transferring or native system of images transferring (iTunes version 11.0), that allowed image visualization. The devices were the following: • 15.6-inch notebook LCD display (ASUS G51Jx, ASUSTeK Computer Inc., Taipei City, Taiwan) with full high definition resolution of 1920 9 1080 pixels. Software: Photoshop Lightroom 5.4 (Adobe Systems, San Jose, CA, USA). • 19-inch desktop LCD display (LG Flatron L1953HSF, LG Electronics, Seoul, Korea) with a resolution

(a)

of 1024 9 768 pixels. Software: Photoshop Lightroom 5.4. (Adobe Systems). • 8-inch AndroidTM tablet (Galaxy Note 8.0, Samsung, Seoul, Korea) with an LCD TFT screen with high definition resolution of 1280 9 800 pixels. Software: Photoshop Express v. 2.2.190 (Adobe Systems). • 9.7-inch iPadâ fourth-generation tablet (Apple Inc., Cupertino, CA, USA) with an LCD IPS screen with retina resolution of 2048 9 1536 pixels. Software: Photoshop Express v. 3.2. (Adobe Systems). The screens of all devices were set and fixed to their maximum brightness. Different softwares were used for each computer and tablet because they use different operating systems, and software that runs on all of them is not available. Nevertheless, the software used was all from the same company and had the same features needed for this study (open image files, zoom and change brightness and contrast). The set of three images for each root was evaluated blindly by 3 calibrated oral and maxillofacial radiologists, under dim light conditions. A calibration procedure was undertaken with all evaluators to standardize the VFR concept and diagnostic. Despite the portable aspect of most devices used, the evaluations were conducted in the same room, to standardize the viewing conditions. The observers could adjust image brightness and contrast. Each radiograph was evaluated twice at an interval of at least 4 weeks.

(b)

Figure 2 Periapical images showing the vertical root fracture (a) without root canal filling (arrow) and (b) with a fibreglass

post.

388

International Endodontic Journal, 49, 386–392, 2016

© 2015 International Endodontic Journal. Published by John Wiley & Sons Ltd

Vasconcelos et al. Digital radiographs on different devices

The observers recorded their observations on a 5point scale as follows: 1, fracture definitely not present; 2, fracture probably not present; 3, unsure; 4, fracture probably present; and 5, fracture definitely present. The data were analysed using SAS (9.1.3 version; SAS Institute Inc., Cary, NC, USA). Intra- and interobserver agreements were calculated using the weighted kappa test. The sensitivity, specificity and accuracy were calculated. Receiver operating characteristic (ROC) analyses were performed to compare the evaluations with the gold standard. The values for the sensitivity, specificity, accuracy and areas under the ROC curve (Az) for each observer, each device and under different conditions were compared by two-way analysis of variance with the post hoc Tukey test. A significance level of 5% was used for all analyses.

Discussion Radiographic assessment is important in the diagnosis of VRFs (Tofangchiha et al. 2013). However, the quality of digital radiographic images depends on each component of the imaging procedure being maintained at a high quality (Hellen-Halme et al. 2007). According to Samei (2003), the monitors seem to be the weakest point in the process. For this reason and the increasing use of mobile devices, this study aimed to compare 4 devices/displays with different resolutions and screen sizes for root fracture diagnosis. Then, a condition similar to a clinical reality was simulated as most teeth with VRFs have root fillings or posts that is a limiting factor for diagnosis. Although there are various types of post, only one type was used to avoid the influence of other variable factors. The diagnostic accuracy for root fracture detection was not significantly different between the displays/devices. The accuracy values are similar to previous studies in which conventional monitors were used (Khedmat et al. 2012, da Silveira et al. 2013). The intra- and interobserver reproducibility results also follow the general trend from previous studies (Hassan et al. 2009, Kamburo glu et al. 2009, Varshosaz et al. 2010, Tofangchiha et al. 2011, 2013, Kambungton et al. 2012). Only one study was found that tested the accuracy of 3 monitors (2 medical and one conventional) for detecting root fracture, and the authors concluded that the type of monitor did not influence the diagnosis of VRFs (Tofangchiha et al. 2013). However, there are 3 main differences between that study and the present one. First, an intracanal post was used. Moreover, 4 different devices/displays were used, including mobile devices with smaller screen size. In addition, the evaluation in the current study was performed by 3 observers, for all devices, which provide more reliable results. As long as all observers have the same experience and are well calibrated, a larger number of

Results The weighted kappa coefficients for intra- and interobserver reproducibility are given in Table 1. Intraobserver reproducibility ranged from moderate to almost perfect agreement and interobserver reproducibility ranged from fair to substantial agreement (Landis & Koch 1977). Moreover, they were similar for all the devices. The mean diagnostic values and Az are shown in Table 2. Higher values of sensitivity, accuracy and Az were obtained for root fracture detection on images of unfilled canal; however, only sensitivity and Az values revealed a significant difference when compared with images of teeth with a fibreglass post, independent of the device used; accuracy was not affected by the presence of a post. The values for specificity were higher for images with a fibreglass post, also regardless of the device used, and were significantly different compared with the values for unfilled teeth. On the other hand, there were no significant differences between the devices.

Table 1 Intra- and interobserver reproducibility, according to the weighted Kappa test 15.6-inch full HD notebook

Ex1 Ex2 Ex3

19-inch standard desktop monitor

8-inch Android tablet

9.7-inch iPad Air

Ex 1

Ex 2

Ex 3

Ex 1

Ex 2

Ex 3

Ex 1

Ex 2

Ex 3

Ex 1

Ex 2

Ex 3

0.83

0.43 0.70

0.49 0.35 0.88

0.72

0.47 0.75

0.50 0.40 0.76

0.56

0.35 0.55

0.46 0.31 0.60

0.59

0.59 0.78

0.66 0.57 0.74

Ex, examiner; HD, high definition.

© 2015 International Endodontic Journal. Published by John Wiley & Sons Ltd

International Endodontic Journal, 49, 386–392, 2016

389

Digital radiographs on different devices Vasconcelos et al.

Table 2 Mean (standard deviation) diagnostic values and the area under the ROC curve (Az) for the different devices and con-

ditions

Sensitivity Without post With post P (intracanal condition) Specificity Without post With post P (intracanal condition) Accuracy Without post With post P (intracanal condition) Az Without post With post P (intracanal condition)

15.6-inch full HD notebook

19-inch standard desktop monitor

8-inch Android tablet

9.7-inch iPad Air

P (device)

0.90 (0.05) 0.53 (0.18) 0.000

0.90 (0.09) 0.58 (0.06)

0.82 (0.13) 0.52 (0.18)

0.85 (0.09) 0.48 (0.14)

0.645

0.48 (0.25) 0.85 (0.18) 0.032

0.63 (0.18) 0.78 (0.20)

0.62 (0.15) 0.70 (0.26)

0.63 (0.23) 0.80 (0.09)

0.942

0.70 (0.12) 0.69 (0.08) 0.06

0.77 (0.09) 0.68 (0.12)

0.72 (0.01) 0.62 (0.09)

0.74 (0.08) 0.64 (0.03)

0.708

0.88 (0.03) 0.71 (0.08) 0.000

0.88 (0.05) 0.78 (0.12)

0.82 (0.03) 0.64 (0.04)

0.89 (0.03) 0.71 (0.12)

0.136

HD, high definition.

observers are not necessary, considering that the major aim was not to analyse evaluator performance, but the effect on the devices/displays in the visualization and diagnosis of VRFs. Differences in relation to the area under the ROC curve, sensitivity and specificity were observed when the conditions with and without a post were compared. Nevertheless, the accuracy was not influenced by the presence of the post, in agreement with other studies (Khedmat et al. 2012). The differences in relation to the sensitivity values can be justified because of the characteristics of the specimens where a post was used, which overlaps and masks the fracture line, and poses a challenge in the diagnose of VFR and reduces the sensitivity. This aspect also naturally increases the specificity, as the nondetected fracture cases will be larger than the detected ones. Other studies have evaluated whether the monitor influences the diagnosis of other dental diseases, and no significant differences were found between the devices (Ludlow & Abreu 1999, Esmaeili et al. 2007, Hellen-Halme et al. 2007, Ilg€ uy et al. 2009, Isidor et al. 2009). However, these studies used only conventional monitors. The factors that determine monitor fidelity include the resolution (matrix size), bit depth, dot pitch, luminance and display size. In addition to the monitor fidelity, another important factor that must be considered is observer performance, which may also be

390

International Endodontic Journal, 49, 386–392, 2016

limited by the capacity of the human visual system (White & Pharoah 2013). The monitors commonly used with most computer systems have a nominal resolution ranging from 640 9 480 to 1600 9 1200 pixels. Four different screens were used to evaluate the images in this study. The images were acquired with 25.6 lp mm 1 spatial resolution; therefore, the minimal resolution of all the monitors was sufficient to reveal the information contained in the image, which can explain the fact that no significant differences between them were found. Line pairs per millimetre are measured as the number of white and black lines that fit within 1 mm. Previous studies have shown no significant differences when a high lp mm 1 was used compared to lower lp mm 1 (Li et al. 2008, Oliveira et al. 2012). In addition, K€ unzel et al. (2003) found that the human eye can only see 11.1 lp mm 1 with unaided vision, but that this value increases when digital magnification is used. In the current study, observers were allowed to use the visualization software magnification tools. For that reason, the highest theoretical lp mm 1 the hardware could scan was used to prevent problems that could be caused by the magnification of an image with a lower spatial resolution. The size of the screens could have influenced the evaluations; the mobile devices were half the size of the conventional monitors. However, the results revealed that screen size did not affect the diagnosis of VRFs.

© 2015 International Endodontic Journal. Published by John Wiley & Sons Ltd

Vasconcelos et al. Digital radiographs on different devices

The greyscale in a radiograph is dependent on the bit depth used for acquisition and display. Larger bit depth allows the use of more shades of grey, and the distinction between the different densities of the object is enhanced. Monitors with a bit depth of 16 are capable of displaying 6500 colours, but grey levels are limited for radiographic images (Esmaeili et al. 2007). Moreover, as the images in this study were exported from the acquisition software in PNG format which uses a 8-bit scale, the greyscale presented by all devices was the same; this could explain why the screen with the largest amount of bits, the iPad with 64 bits, did not provide better accuracy. All the devices had more than 8 bits; therefore, all were able to expose the 256 greyscales contained in the image. The American Association of Physicists in Medicine recommends an illuminance of

Digital radiographs displayed on different devices: effect on the detection of vertical root fractures.

To evaluate whether the type of display device affects the detection of vertical root fractures (VRFs) on digital radiographs in unfilled canals and c...
190KB Sizes 0 Downloads 12 Views