Dental magnetic resonance imaging: making the invisible visible

Abstract

Introduction: Clinical dentistry is in need of noninvasive and accurate diagnostic methods to better evaluate dental pathosis. The purpose of this work was to assess the feasibility of a recently developed magnetic resonance imaging (MRI) technique, called SWeep Imaging with Fourier Transform (SWIFT), to visualize dental tissues.

Methods: Three in vitro teeth, representing a limited range of clinical conditions of interest, imaged using a 9.4T system with scanning times ranging from 100 seconds to 25 minutes. In vivo imaging of a subject was performed using a 4T system with a 10-minute scanning time. SWIFT images were compared with traditional two-dimensional radiographs, three-dimensional cone-beam computed tomography (CBCT) scanning, gradient-echo MRI technique, and histological sections.

Results: A resolution of 100 μm was obtained from in vitro teeth. SWIFT also identified the presence and extent of dental caries and fine structures of the teeth, including cracks and accessory canals, which are not visible with existing clinical radiography techniques. Intraoral positioning of the radiofrequency coil produced initial images of multiple adjacent teeth at a resolution of 400 μm.

Conclusions: SWIFT MRI offers simultaneous three-dimensional hard- and soft-tissue imaging of teeth without the use of ionizing radiation. Furthermore, it has the potential to image minute dental structures within clinically relevant scanning times. This technology has implications for endodontists because it offers a potential method to longitudinally evaluate teeth where pulp and root structures have been regenerated.

Figure 1

A mandibular right second molar (Tooth 1) with obvious interproximal and questionable occlusal caries, imaged with different modalities. Two SWIFT images with different scanning times (24 minutes and 100 seconds) are presented. The image acquired in 100 seconds has more noise, but the tooth structure and dental caries are still well recognizable. The application of an additional filtering step is a potential way to increase the SNR ratio and thus produce satisfactory quality images at shorter scanning times. (SWIFT: selected slices with FOV diameter 25 mm and isotropic voxel size 98 μm; CBCT: isotropic voxel size 200 μm)

Figure 2

Three cross-sectional slices of Tooth 1 (marked as S1a, S1b and S1c in Figure 1) are presented. Graphs located on the top of figure with arbitrary gray value present the profiles of slice S1a at the position depicted with a light blue line. The colored arrows delineate the surface of the tooth enamel (red), sound dentin (green) and the signal level of caries (dark blue), respectively. Note that the signal from enamel tissue in SWIFT image is clearly greater than background noise (black arrows). The presence of occlusal caries is observed in S1a within the SWIFT and non-decalcified histology sections, which is the gold-standard measure, but not observed in the CBCT scan. The extent of the interproximal caries towards the dental pulp, in S1b and S1c, are more completely delineated in the SWIFT sections than the CBCT when compared to histology. The slice thickness is equal to 98 μm for SWIFT and 200 μm for CBCT images.

Figure 3

A maxillary right central incisor (Tooth 2) with bilateral interproximal composite resin restorations and recurrent caries, imaged with different modalities. (SWIFT: selected slices with FOV diameter 40 mm and isotropic voxel size 156 μm; CBCT: isotropic voxel size 200 μm).

Figure 4

Two cross-sectional slices of Tooth 2 (marked as S3a and S3b in Figure 3) are presented. The presence of an accessory canal (yellow arrows) is observed in the slice S3a within the SWIFT image and non-decalcified histology section, but not observed in the CBCT section. In slice S3b, the extent of the composite resin restorations are identifiable within the SWIFT section as well as in the CBCT and histological sections. The slice thickness is equal to 156 μm for SWIFT and 200 μm for CBCT images.

Figure 5

A maxillary left first premolar with a complete lingual cusp fracture that has been reapproximated. In both radiographic image types and optical images the crack (yellow arrows) is hard to identify, but is easily observed in the SWIFT images. The red arrow delineates what is most likely air entrapped in the pulp canal when the crack was induced.

Figure 6

Schematic illustration of the position of intra-oral RF coil for in vivo dental imaging experiments on the top of selected slice of a SWIFT image.

Figure 7

In vivo images of the right posterior teeth. The photograph depicts the maxillary teeth that are also imaged with a traditional 2D radiograph used to detect interproximal caries. The dotted lines, represented by a, b, c and d, correlate with the cross-sectional CBCT and SWIFT images at those levels, from more superior closer to the root tip moving inferiorly to the crown of the teeth. Note the lack of image distortion associated with the occlusal amalgam restorations in the SWIFT sections compared to the CBCT sections. (SWIFT: selected slices with FOV diameter 110 mm and isotropic voxel size 430 μm).