Validity of diagnosis of questionable caries lesions in occlusal surfaces of extracted third molars

This study was to evaluate the accuracy of visual inspection, conventional radiography, and digital radiographic methods for the detection of occlusal carious lesions in third molars from a present-day adolescent population. Seventy-eight third molars, considered clinically to be without occlusal cavities, were extracted from young soldiers. Before extraction, an intra-oral radiograph was obtained. After extraction, the teeth were examined visually as per the criteria: 0 = no caries, 1 = chalky/stained fissure indicative of enamel caries, 2 = chalky and dark-stained fissure considered indicative of a dentinal lesion, and 3 = as per criterion 2, but with small surface defects (microcavities). The radiographs were digitised, and image enhancement with contrast stretch and a filtering procedure was performed, respectively. The three types of radiographic image were assessed as per the criteria: 1 = no caries/caries confined to enamel, 2 = caries involving the outermost dentine, and 3 = deep dentinal caries extending half-way or more to the pulp. Ground sections (500-600 microns in thickness) served as validation for lesion depth. The digital radiographic method with contrast stretch performed overall best of the four methods (greater than 70% detection rate) while visual inspection (53% detection rate) performed better than conventional radiography (48% detection rate). When results from visual inspection and conventional radiography were combined, an increase in the detection rate of 11% was obtained with a 7% increase in false-positive scorings. When digitally contrast-manipulated radiographs were combined with visual inspection, a gain of 33% was obtained with an 11% increase in false positives.(ABSTRACT TRUNCATED AT 250 WORDS)