Detecting Early Carious Lesions: An Ex-vivo MRI-based Study

Objectives: The diagnostic accuracy of early carious lesions is less than desirable, largely due to a low sensitivity to detect shallow depth lesions with clinically significant mineral loss. The aim of this study was to determine the diagnostic accuracy of MRI based system, compared to microCT, for early caries detection. Methods: Fifteen early carious lesions were placed in 15 tooth samples with remineralization attempted in 10. All samples were stored in isotopic saline and imaged after being patted dry. The index test was MRI, performed using 9.4T Varian system with DirectDrive and a custom fabricated radiofrequency coil. The reference standard was microCT (XT H 225, Metris). To determine whether a carious lesion was present (yes/no), two blinded observers derived consensus agreement on lesions status. Calculations of sensitivity and specificity were used to determine diagnostic accuracy. In addition, test-retest reliability was determined using the kappa statistics. MRI based lesion status was determined at two different time points separated by 6 days. The magnitude of kappa was compared with guidelines suggesting values between 0.8 and 1.00 as “excellent”. Results: The sensitivity was 80% (95% CI: 44-98%) and specificity was 80% (95% CI: 28-100%). Test-retest reliability was “excellent”, indicated by a kappa=0.86 (95% CI: 0.61-1.00). Conclusion: Preliminary ex-vivo results of MRI based detection of early carious lesions suggest acceptable diagnostic accuracy. Given the enhanced contrast MRI has over x-ray based techniques, since the relative change of water content is greater than mineral content in densely calcified tissues, research assessing in-vivo use is warranted to explore potential for develop for clinical applications. Supported by the NIH Grants P41-RR008079, K12-RR023247, S10-RR023730 and by the Keck Foundation.