Assessing the Accuracy and Reliability of Root Crack and Fracture Detection in Teeth Using Sweep Imaging with Fourier Transform (SWIFT) Magnetic Resonance Imaging (MRI)

Abstract

Introduction: Magnetic Resonance Imaging (MRI) has the potential to aid in determining the presence and extent of cracks/fractures in teeth due to more advantageous contrast, without ionizing radiation. An MRI technique called Sweep Imaging with Fourier Transform (SWIFT) has overcome many of the inherent difficulties of conventional MRI with detecting fast-relaxing signals from densely mineralized dental tissues. The objectives of this in vitro investigation were to develop MRI criteria for root crack/fracture identification in teeth and to establish intra- and inter-rater reliabilities and corresponding sensitivity and specificity values for the detection of tooth-root cracks/fractures in SWIFT MRI and limited field of view (FOV) CBCT. Materials and Methods: MRI-based criteria for crack/fracture appearance was developed by an MRI physicist and 6 dentists, including 3 endodontists and 1 Oral and Maxillofacial (OMF) radiologist. Twenty-nine human adult teeth previously extracted following clinical diagnosis by a board-certified endodontist of a root crack/fracture were frequency-matched to 29 non-cracked controls. Crack/fracture status confirmation was performed with magnified visual inspection, transillumination and vital staining. Samples were scanned with two 3D imaging modalities: 1) SWIFT MRI (10 teeth/scan) via a custom oral radiofrequency (RF) coil and a 90cm, 4-T magnet; 2) Limited FOV CBCT (1 tooth/scan) via a Carestream (CS) 9000 (Rochester, NY). Following a training period, a blinded 4-member panel (3 endodontists, 1 OMF radiologist) evaluated the images with a proportion randomly re-tested to establish intra-rater reliability. Overall observer agreement was measured using Cohen’s kappa and levels of agreement judged using the criteria of Landis and Koch. Sensitivity and specificity were computed with 95% confidence interval (CI); statistical significance was set at 0.05. Results: MRI-based crack/fracture criteria were defined as 1-2 sharply-delineated, high-signal (bright/white) line shape(s) that must be visible on multiple contiguous image slices. The line shape(s) must present as: single entities, or parallel pairs in close proximity, or pairs in close proximity exhibiting convergence or divergence extending from the external boundary of the tooth to the pulpal cavity. Intra-rater reliability for MRI was fair-to-almost perfect (κ=0.38-1.00) and for CBCT was moderate-to-almost perfect (κ=0.66-1.00). Inter-rater reliability for MRI was fair (κ=0.21; 95%CI:0.10-0.31; p< 0.001) and for CBCT was moderate (κ=0.45; 95%CI:0.34-0.56; p<0.001). Sensitivity: MRI=0.59 (95% CI:0.39-0.76; p=0.46); CBCT=0.59 (95% CI:0.59-0.76; p=0.46). Specificity: MRI=0.83 (95% CI:0.64-0.94; p<0.01); CBCT=0.90 (95% CI:0.73-0.98; p<0.01). Conclusions: Education and training for both imaging modalities is needed to improve reliabilities for the identification of tooth-root crack/fractures. Despite the advantages of increased contrast and absence of artifact from radio-dense materials in MRI, comparable measures of sensitivity and specificity (in relation to CBCT) suggest quality MRI improvements are needed, specifically in image acquisition and post-processing parameters. Given the early stage of technology development and multiple available pathways to optimize MR imaging of teeth, there may be a use for SWIFT MRI in detecting cracks and fractures in teeth.