The present study had an in-vitro design and was conducted after acceptance by the local Institutional Review Board (IRB) (protocol number #4.231.578).
Sample selection and preparation
The sample was composed of 34 single-rooted human teeth, including lower incisors, canines and premolars and divided into the control group (without VRF; n = 15 teeth) and the experimental group (with VRF; n = 19 teeth). All teeth were collected from a private dental clinic after being extracted mainly due to the presence of periodontal disease, extensive coronal caries lesions or orthodontic indication. All teeth were disinfected in 2% glutaraldehyde and submitted to clinical and radiographic evaluation to ensure the absence of dental anomalies, calcification, endodontic treatment, open apex, resorptive lesions, and fractures. All teeth were sectioned at the level of the cementoenamel junction to avoid memorization of tooth morphology and bias related to coronary fracture identification during the evaluation. Then, vertical fractures were induced in each of 19 roots (experimental group), using a universal testing machine (INSTRON 4411, Instron Corporation, Canton, MA) and following a previously established methodology . Transillumination using a LED light (Ultralume 5, Ultradent Products Inc., South Jordan, UT, USA) was used to verify the presence of the VRFs, which were located at least at the cervical and middle third of the root.
Image acquisition and export
The roots were individually included in alveolar sockets of a sectioned human mandible for periapical radiography acquisition. Also, to simulate a clinical scenario, a tongue made of wax was placed behind the image receptor .
With the aid of a custom-made acrylic apparatus, the radiographic images were acquired following the parallelism technique. This apparatus was composed of a locator ring to standardize the position of the position-indicating device of the X-ray unit (distance focus-receptor standardized in 40 cm, horizontal angulation of 90º, and vertical angulation of 0º). Further, the apparatus allowed to maintain the distance between the object and the receptor in 1.5 cm. To reproduce a clinical scenario and mimic the attenuation of the X-rays by the soft tissues, an acrylic block of 2.5 cm in thickness was settled in front of the buccal cortical plate of the mandible.
The periapical radiographic images were acquired using two radiographic systems with distinct technologies:
Phosphor plate (PSP): Digora Optime (Soredex, Tuusula, Finland), size 2, 8-bit contrast resolution, and theoretical spatial resolution of 14.3 lp mm−1.
CMOS sensor: Digora Toto—Scanora (Soredex, Tuusula, Finland), size 2, 12-bit contrast resolution, and theoretical spatial resolution of 26.3 lp mm−1.
All images were obtained using the same X-ray unit (FocusTM—Instrumentarium Dental Inc., Milwaukee, WI, USA), under the same acquisition protocol: 60 kVp, 7 mA, exposure time as recommended by the receptors’ manufacturers (0.3 s for Digora Optime, and 0.18 s for Digora Toto), and standardized focus-receptor distance of 40 cm.
After radiographic acquisition, the images were individually exported from their native software into the four different file formats—TIFF, BMP, PNG, and JPEG—, totaling 272 radiographs (34 radiographs × 2 digital systems × 4 file formats). The Table 1 shows the average file size of each radiographic image for both systems studied, according to the experimental conditions of the present study. Also, Figs. 1 and 2 show radiographs of a sound and of a fractured root, respectively, obtained with the two digital systems tested and exported into the four different file formats.
Images were independently evaluated by five oral and maxillofacial radiologists with five years of experience in radiographic diagnosis, who were blinded concerning the experimental design, in a medical display (MDRC-2124, Barco N.V., Courtray, Belgium) with 1920 × 1200 pixels resolution. The evaluations were conducted using the JPEGView image viewer. Before the evaluations, the examiners were instructed about the assessment of VRFs, using periapical radiographs that were not included in the sample as examples. To simulating a clinical scenario, they were also allowed to adjust image brightness, contrast, and zoom settings. Then, examiners should classify each image according to the presence or absence of VRF, using a 5-point scale: 1—definitely absent; 2—probably absent; 3—uncertain; 4—probably present; 5—definitely present. To reduce the possibility of image memorization and prevent visual weariness, the evaluators were oriented to assess a maximum of 20 images per day. Thirty days after the analysis was concluded, 30% of the sample was re-evaluated to calculate the intra-examiner agreement.
All evaluators were blind for the image file format, as all images were codified and randomized to prevent identification.
Data were analyzed in the Statistical Package for Social Sciences software v. 24.0 (IBM Corp., Armonk, NY, USA). Intra- and inter-examiner agreements regarding the diagnosis of VRF were calculated by weighted Kappa test. Diagnostic values [area under the receiver operating characteristic (ROC) curve, specificity, and sensitivity] for the diagnosis of VRF were calculated. The data obtained were individually compared by two-way analysis of variance with Tukey’s test as a post-hoc, considering the file formats and radiographic systems as the studied factors. A significance level of 5% was adopted for all analyses.