RESUMO
PURPOSE: The aim of this study was to compare the performance of direct digital radiography and traditional dental radiography for the detection of proximal surface dental caries in the mixed dentition. METHODS: 15 quadrants of extracted teeth, arranged from the primary canine to permanent first molar, were imaged using direct digital (Schick Technologies, Long Island City, NY, USA) and conventional films (D-speed and E-speed Plus; Eastman Kodak Co., Rochester, NY, USA). Five pediatric dentists viewed the images and scored the 270 proximal surfaces for presence of caries on a 5 point scale and extent of caries on a 4 point scale. The teeth were sectioned and viewed microscopically to determine the gold standard. Receiver operating characteristic (ROC) analysis and analysis of variance (ANOVA) were used to evaluate the viewer's performance for detecting proximal caries using the 3 different image receptor types. RESULTS: Experienced examiners were significantly more accurate in diagnosis of proximal surface caries using either D-speed or E-speed Plus films than they were using the direct digital receptor. The mean areas under the ROC curve (Az) for the viewers were 0.7595 for D-speed film, 0.7557 for E-speed Plus film, and 0.5928 for the direct digital receptor. The results also indicated that selected viewers' accuracy increased when viewing the direct digital images a second time. CONCLUSION: CCD based direct digital radiography was not as accurate as conventional film images for the purpose of diagnosing proximal surface caries in the mixed dentition. However, the results imply that with increased experience, direct digital images may be as accurate as conventional film based images for diagnosis.