Browsing by Author "Liang, Tracy"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Reproducibility, Repeatability, and Precision: Digital Dental Model Registration with VIBE, T2 Space, and MP-RAGE Magnetic Resonance Imaging
    (2022-06) Liang, Tracy
    PURPOSE: Multi-modal registration of digital dental models and cone beam computed tomography (cbCT) volumes provides augmented 3-D information for comprehensive orthodontic diagnosis, treatment planning, surgical prediction, growth assessment, and outcome evaluation. This study compared reliability and precision of surface-based registration of a digital dental model to that of registration with magnetic resonance image volumes acquired using 3-D volumetric interpolated breath-hold examination (VIBE), T2 sampling perfection with application optimized contrasts using different flip angle evolution (SPACE), and magnetization-prepared rapid gradient-echo (MP-RAGE) pulse sequences to assess whether MRI may be a suitable replacement to cbCT. Secondarily, this study examined what effect intraoral metal alloy fixtures have on precision.DESIGN: A cross-sectional diagnostic study. SUBJECTS: CbCT and digital dental model data were previously acquired from eight adult patients who sought orthodontic treatment at the University of Minnesota. Following recruitment, MRI scans were acquired from these same individuals at the Center for Magnetic Resonance Imaging. METHODS: CbCT and all three MRI volume types were segmented semi-automatically and saved in stereolithographic (STL) format. Digital model registration with cbCT, T2 Space, VIBE, and MP-RAGE volumes were completed manually to achieve best fit. Then, automatic surface-based registration was initiated using incisal or occlusal and facial surfaces of the maxillary dentition. To evaluate the reliability and precision of the process, differences between the two registered surfaces were measured at the first molars and central incisors. Registrations were repeated 88 times by three operators for cbCT, T2 Space, VIBE, and MP-RAGE volumes. Re-test registrations were then repeated 33 times in the same manner. PRIMARY OUTCOME MEASURE: Root mean square (RMS) error in millimeters (mm) measured by 3dMD Vultus software. Clinical acceptability was designated as ≤ 0.5 mm RMS error. RESULTS: Inter-operator reproducibility of registrations with cbCT, T2 Space, VIBE, and MP-RAGE (i.e. imaging modalities) measured by RMS error showed no statistically significant difference among the three operators for each imaging modality (p=0.0771, F=3.09). Re-test values were statistically equivalent to the primary dataset for cbCT, T2 Space, VIBE and MP-RAGE (p=0.9975, p=1.0000, p=0.9992, p=0.9984), confirming intra-operator repeatability. The variables of imaging modality and interaction of imaging and dental landmark had the most statistically significant influence on precision of registration (F=178.23, F=63.34 with p values <.0001). The overall precision of registration for cbCT, T2 Space, VIBE, and MP-RAGE was 0.338 mm (0.316, 0.361), 0.437 mm (0.419, 0.455), 0.627 mm (0.599, 0.655), and 0.532 mm (0.512, 0.552), respectively. The RMS error values of molars was significantly different from incisors for cbCT, VIBE, and MP-RAGE (p<.0001, p<.0001, and p=0.0021). For incisors RMS error, cbCT was significantly lower than T2 Space, VIBE, and MP-RAGE (p<.0001). For molars RMS error, cbCT was not statistically different from T2 Space (p=1.0000) and statistically lower than VIBE (p=0.0003) and MP-RAGE (p=0.0295). Bland-Altman analysis demonstrated excellent agreement when evaluating reproducibility and precision for each landmark-modality combination. In participants with no posterior metal artifacts, molars RMS error for cbCT was significantly lower than for T2 Space (p=0.0031), VIBE (p=0.0036) and MP-RAGE (p=0.0041); incisors RMS error for cbCT was not statistically different from T2 Space (p=0.1765) and statistically lower than VIBE (p=0.0162) and MP-RAGE (p=0.0056). In participants with posterior metal artifacts, molars RMS error for cbCT was statistically higher than T2 Space (p=0.0043), VIBE (0.0110), and MP-RAGE (p=0.0056); incisors RMS error for cbCT was statistically lower than T2 Space (p=0.0079), VIBE (p=0.0032) and MP-RAGE (p=0.0043). CONCLUSION: Using the proposed method, the results demonstrated reproducibility and repeatability of regional, surface-based registration of digital dental model and MRI acquired using T2 Space, 3-D VIBE, and MP-RAGE pulse sequences. Overall precision of cbCT registrations was statistically higher than that of VIBE and MP-RAGE. However, cbCT registrations precision was statistically higher than T2 Space for incisors only; there was no statistically significant difference with respect to molars. The interaction of imaging modality and landmark had a significant effect on precision, most likely due to presence of posterior artifact intraorally. Presence of posterior metal artifacts had an effect on precision of molars RMS for cbCT. Air artifacts confounded analysis of incisors RMS error for T2 Space, VIBE, and MP-RAGE. Image distortion from posterior metal artifact was more severe in cbCT than T2 Space, VIBE, or MP-RAGE, with T2 Space registrations exhibiting the lowest RMS error, though only 65% of all RMS error values were clinically acceptable, <0.5 mm. Further optimization of the T2 pulse sequence acquisition parameters can improve precision of the surface-based registration method for MRI use in the general population.