Narrow Band Imaging by Novice Raters After Short Training: A Pilot Feasibility Study of Diagnostic Accuracy and Repeatability

OBJECTIVE: To pilot the feasibility, diagnostic accuracy, and repeatability of the Takano NBI-based IPCL classification when applied by novice raters after short formal training. METHODS: Ten dental students classified 90 oral lesions shown as paired white-light and NBI images in two sessions, 4 weeks apart. Scores were recorded on an ordinal scale (0-4), where category 0 denoted hyperplasia or clinically normal mucosa, while categories 1-4 represented increasing degrees of epithelial dysplasia or invasive disease. Histopathology served as the reference standard. RESULTS: The 90 lesions comprised 38 hyperplastic/normal, 29 dysplastic, and 23 invasive carcinomas. Mean agreement with histology was 18.8% at T1 and 21.0% at T2, with weighted kappa values of 0.080 and 0.093, respectively. Agreement with the expert rater was 22.2% at T1 and 20.4% at T2, with weighted kappa values of 0.170 and 0.127. CONCLUSION: A short lecture-based training is insufficient for reliable NBI interpretation by novice raters. These findings support the necessity for structured, competency-based training and supervised case libraries before Takano-based NBI assessment can be recommended in clinical practice for inexperienced practitioners.