| Eggertsson H, Ando M, Ferreira-Zandoná AG, González-Cabezas C, Fontana MR, Martinez-Mier EA, Jackson RD, Krushinski CA, Crawford J, Morgan DS, Stookey GK,Zero DT (2002) Variation of image acquisition and image analysis using QLF in vitro.. Caries Res. 36: p. 191 |
|
| |
Quantitative light-induced fluorescence (QLF) has been shown to be an effective method for early caries detection. Two sources of variation in using the method come from the image acquisition and from analysis of lesions from the images. This study assessed the impact of each source of variation on the outcome of lesion analysis. Several examiners were trained in using the QLF equipment for imaging and the specialized software for the image analysis. Ten non-cavitated natural lesions on extracted teeth were selected: 5 smooth surface lesions and 5 occlusal lesions. The study was fourfold. Part 1: Seven examiners took images and analyzed their own images. Part 2: One examiner analyzed the images taken by all seven examiners. Part 3: One examiner analyzed images taken by one examiner 5 times. Part 4: Six examiners analyzed images taken by one examiner. Lesions were analyzed for average loss of fluorescence, area and ˘Q. Agreement within and between examiners was examined using intraclass correlation coefficients (ICC). Results (based on ˘Q) were: Part 1: ICC for combined variability between examiners from taking and evaluating images was 0.88 and 0.94 for occlusal and smooth surfaces, respectively. Part 2: When one analyzer was used for these images, ICC was 0.91 and 0.95. Part 3: Repeated analysis of images taken by one examiner, ICC was 0.99 for both types of lesions. Part 4: Variability of using different examiners analyzing images taken by one examiner, ICC was 0.99 for occlusal and 0.98 for smooth surface lesions. In conclusion, the overall variation using QLF was low. Image acquisition by different examiners was a greater source of variation than analysis done by different examiners on images taken by one examiner. |