Endodontic working length measurements of premolars and molars in high-resolution dental MRI: a clinical pilot study for assessment of reliability and accuracy

Abstract

ObjectivesTo prospectively assess the reliability and accuracy of high-resolution, dental MRI (dMRI) for endodontic working length (WL) measurements of premolars and molars under clinical conditions.Materials and methodsThree-Tesla dMRI was performed in 9 subjects who also had undergone cone-beam computed tomography (CBCT) (mean age: 47 ± 13.5 years). A total of 34 root canals from 12 molars (4/8, upper/lower jaw; 22 root canals) and 11 premolars (2/9 upper/lower jaw; 12 root canals) were included. CBCT and dMRI datasets were reconstructed to visualize the root canal in one single slice. Subsequently, two radiologists measured the root canal lengths in both modalities twice in blinded fashion. Reliability and accuracy for both modalities were assessed using intraclass correlation coefficients (ICCs) and Bland–Altman analysis, respectively.ResultsReliability (intra-rater I/II; inter-rater) of dental MRI measurements was excellent and comparable to CBCT for premolars (0.993/0.900; 0.958 vs. 0.993/0.956; 0.951) and for molars (0.978/0.995; 0.986 vs. 0.992/0.996; 0.989). Bland–Altman analysis revealed a mean underestimation/bias (95% confidence interval) of dMRI measurements of 0.8 (− 1.44/3.05) mm for premolars and 0.4 (− 1.55/2.39) mm for molars. In up to 59% of the cases, the accuracy of dMRI for WL measurements was within the underestimation margin of 0 to 2 mm short of the apical foramen AF.ConclusionsIn vivo demonstration and measurement of WL are feasible using dMRI. The reliability of measurements is high and equivalent to CBCT. Nonetheless, due to lower spatial resolution and longer acquisition time, the accuracy of dMRI is inferior to CBCT, impeding its current use for clinical treatment planning.Clinical relevancedMRI is a promising radiation-free imaging technique. Its reliability for endodontic working length measurements is high, but its accuracy is not satisfactory enough yet.