IADR Abstract Archives
Correlation of Occlusal Caries Detection Methods With Histological Depth
Objectives: Evaluate the International Caries Detection and Assessment System (ICDAS) visual criteria and fluorescence-based caries detection devices for the detection of occlusal caries with histology as gold standard.
Methods: Posterior permanent teeth (n=60) with ICDAS scores 0-4 on occlusal surfaces were selected, mounted on manikins and examined on a Phantom-head. Four calibrated examiners independently evaluated a pre-selected site twice using each of the methods: ICDAS, Quantitative Light-induced Fluorescence (QLF), and Spectra. For QLF, each examiner obtained DF (%fluorescence loss) and DQ (%fluorescence loss vs. area) and Darea. A slice containing the site was obtained, ground and imaged at every 100µm (Nikon SMZ 745T, magnification 5x). The most affected image was evaluated for lesion depth (E1, E2, E2 with DEJ involvement, D1, D2, and D3). Association between methods and histological depth was evaluated using two-way contingency tables and Kendall’s tau-b (t-b) correlation coefficients. Spectra and QLF were changed to categorical variables for sensitivity, specificity, accuracy (AC) and area under the curve (AUR) computations. Spectra manufacture’s guidelines were used to select the categorical variables. For QLF a decision tree model (CART method) was used.
Results: Inter- and intra-examiner ICC ranged from 0.78-0.92 and 0.64-0.96 respectively. Correlation between histology and detection methods were as follows: QLF deltaQ t-b=0.70; QLF deltaF t-b=0.69; QLF- delta area t-b=0.65; ICDAS t-b=0.67; Spectra t-b=0.57 (p-value<0.001). Sensitivity was highest for QLF deltaF (0.55), followed by QLF deltaQ (0.52), QLF- delta area (0.49), ICDAS (0.48) and Spectra (0.35). Specificity was highest for QLF- delta area (0.93), followed by Spectra (0.92), QLF deltaQ (0.91), QLF deltaF (0.89), and ICDAS (0.78). QLF- delta area had the highest accuracy (0.90) and AUC (0.93), followed by QLF deltaQ (0.89 & 0.92); QLF deltaF (0.80 & 0.91); Spectra (0.77 & 0.84) and ICDAS (0.70 & 0.86).
Conclusions: All methods tested were able to detect occlusal caries with 70% or better accuracy, however fluorescence based methods performed better.
Methods: Posterior permanent teeth (n=60) with ICDAS scores 0-4 on occlusal surfaces were selected, mounted on manikins and examined on a Phantom-head. Four calibrated examiners independently evaluated a pre-selected site twice using each of the methods: ICDAS, Quantitative Light-induced Fluorescence (QLF), and Spectra. For QLF, each examiner obtained DF (%fluorescence loss) and DQ (%fluorescence loss vs. area) and Darea. A slice containing the site was obtained, ground and imaged at every 100µm (Nikon SMZ 745T, magnification 5x). The most affected image was evaluated for lesion depth (E1, E2, E2 with DEJ involvement, D1, D2, and D3). Association between methods and histological depth was evaluated using two-way contingency tables and Kendall’s tau-b (t-b) correlation coefficients. Spectra and QLF were changed to categorical variables for sensitivity, specificity, accuracy (AC) and area under the curve (AUR) computations. Spectra manufacture’s guidelines were used to select the categorical variables. For QLF a decision tree model (CART method) was used.
Results: Inter- and intra-examiner ICC ranged from 0.78-0.92 and 0.64-0.96 respectively. Correlation between histology and detection methods were as follows: QLF deltaQ t-b=0.70; QLF deltaF t-b=0.69; QLF- delta area t-b=0.65; ICDAS t-b=0.67; Spectra t-b=0.57 (p-value<0.001). Sensitivity was highest for QLF deltaF (0.55), followed by QLF deltaQ (0.52), QLF- delta area (0.49), ICDAS (0.48) and Spectra (0.35). Specificity was highest for QLF- delta area (0.93), followed by Spectra (0.92), QLF deltaQ (0.91), QLF deltaF (0.89), and ICDAS (0.78). QLF- delta area had the highest accuracy (0.90) and AUC (0.93), followed by QLF deltaQ (0.89 & 0.92); QLF deltaF (0.80 & 0.91); Spectra (0.77 & 0.84) and ICDAS (0.70 & 0.86).
Conclusions: All methods tested were able to detect occlusal caries with 70% or better accuracy, however fluorescence based methods performed better.