Development of a fluorescence-image scoring system for assessing noncavitated occlusal caries

Background This study aimed (1) to develop a scoring system based on a quantitative light-induced fluorescence (QLF) score for the occlusal caries (QS-Occlusal) that standardizes the fluorescence properties of noncavitated lesions from QLF images, (2) to confirm the validity and reliability of QS-Occlusal, and (3) to determine whether it is possible to replace existing clinical examinations by image evaluations based on the developed QS-Occlusal for assessing occlusal caries lesions. Methods This clinical study investigated 791 teeth of 94 subjects. The teeth were assessed by visual and tactile examinations using ICDAS criteria and quantitative light-induced fluorescence-digital (QLF-D) image examinations. QS-Occlusal was divided into four stages (from 0 to 3) based on the progression level of the lesion and the fluorescence loss and red fluorescence on captured QLF-D images. Two trained examiners who were not involved in the visual examination evaluated occlusal fluorescence images using QS-Occlusal. The maximum loss of fluorescence (|ΔF_max|) and the maximum change in the ratio of red and green fluorescence (ΔR_max) were quantitatively analyzed by the QA2 software to detect differences between the QS-Occlusal groups. The modalities were compared in terms of sensitivity, specificity, and area under the receiver operating characteristics (AUROC) curve for three different thresholds of the ICDAS codes: 0 vs 1–4 (D₁), 0–2 vs 3/4 (D₂), and 0–3 vs 4 (D₃). Results |ΔF_max| increased significantly by about 4.7-fold (from 15.94 to 75.63) when QS-Occlusal increased from 0 to 3. ΔR_max was about 6.2-fold higher for QS-Occlusal = 1 (49.74) than for QS-Occlusal = 0 (8.04), and 21.6-fold higher for QS-Occlusal = 3 (P < 0.05). The new QS-Occlusal showed an excellent AUROC (ranging from 0.807 to 0.976) in detecting occlusal caries when optimum cutoff values were applied. The intra- and interexaminer agreements indicated excellent reliability, with ICC values of 0.94 and 0.86, respectively. Conclusions The QS-Occlusal proposed in this study can be used in the clinical detection of noncavitated lesions with an excellent diagnostic ability. This makes it possible to replace clinical examinations and intuitively evaluate the lesion severity and status relatively easily and objectively by applying this scoring system to fluorescence images.