Sample selection
Research protocol was approved by the Research Ethics Committee (PKUSSIRB-201417105). Fifteen extracted mandibular molars with intact mesial roots and closed apices were chosen, based on the power analysis calculation in a previous study by our institution [15]. Briefly, an effect size of 1.93, an alpha-type error of 0.05 and a power beta of 0.95 were input together into an independent samples test from the t-test family (G*Power 3.1.9.3 for Macintosh; Heinrich Heine, Dusseldorf, Germany). The results implied a minimum sample size of 9 samples/group. Taking another canal transportation and centering ability study into consideration [16], a sample size of 15 canals/group was eventually decided.
Teeth with fractures, restorations or caries were excluded. Pre-operative radiographs from different angles (a horizontal angle of 0° and a 30° from the mesial) were taken to make sure all mesial root canals were separated and narrow, with curvatures of 20° to 40° [17].
Pre-operative micro-CT scan
The teeth were decoronated approximately 2 mm above the cemento-enamel junction (CEJ) using diamond burs.
A micro-CT system (vivaCT-40, Scanco Medical, Bassersdorf, Switzerland) at 21 μm nominal voxel size, 56 kVp energy, 142 μA intensity, and 200 ms integration time was used to acquire scans. The region of interest extending from the CEJ to the apex were analyzed and reconstructed by Scanco evaluation software (v6.6, Scanco Medical). Samples were verified that the selected canals were separate and constricted, with initial apical size not larger than #8.
Glide path preparation
A total of 30 canals, 15 mesiobuccal (MB, n = 15) and 15 mesiolingual (ML, n = 15) canals, were selected from the sample teeth and negotiated with #6-#8 stainless steel K-files (Roydent, Johnson City, TN, USA) to achieve apical patency. The working length was determined under an operating microscope by inserting #8 K-file slightly passing the canal terminus and subtracting 0.5 mm from this measurement.
Canals were randomly assigned into two experimental groups (www.random.org) [18]. Briefly, stratification random sampling was applied, with canals being stratified as MB and ML first and followed by simple random sampling. Group 1: Glide path was established by using pre-curved #10-#15 stainless steel K-files (Roydent) manually to the working length; Group 2: Mechanical Glide Path (MGP) super-files #10-#15 (MANI Inc., Japan) were used with OGP setting (OGP 90°, 300 rpm) on Tri-Auto ZX2 cordless motor (J. Morita CORP.). Following the manufacturer’s recommendations, in-and-out pecking movement as well as horizontal reciprocating movement were carefully combined and conducted.
RC-Prep (Premier, Plymouth Meeting, PA, USA) was used as a lubricating agent and the canals were irrigated with copious 3% sodium hypochlorite. Each instrument, manual or mechanical, was used to prepare only 2 canals (as in one mesial root). Any file distortion and separation were observed under microscope and recorded in both groups, and specimens with separated files were excluded at this point. All the preparation was performed by a single dentist with expertise in both preparation techniques.
Post-operative micro-CT scan and image analysis
A total of 30 canals (n = 15 in each group) was scanned post-operatively. To ensure the standardization of the specimens during scanning, silicon rubber discs were used as positioners to keep specimens in the same relative positions for both pre- and post-operative scans, with the same protocol and parameter settings.
3D Slicer 4.6.2 software [19] was used to co-register the pre- and post-operative scans, followed by the application of ITK-SNAP 3.6.0 software to overlap the co-registered images (Fig. 1). Fiji 1.46r software (ImageJ, Madison, WI) was used to do the measurements, so that canal transportation and the centering ratio were calculated at three cross-section levels that corresponded to 1-mm, 3-mm, and 5-mm distance from the root apex by using the equations described by Gambill et al. [20].
Statistical analysis
Statistical analysis was performed with GraphPad Prism 6.2 software (GraphPad Software Inc., San Diego, CA, USA). Shapiro–Wilk normality test was applied to confirm the normality distribution of data (P > 0.05), followed by paired t-test to compare between two groups at different levels. Values of P less than 0.05 were considered statistically significant.