Reliability of landmark recording on film and digital lateral cephalograms

The aims of this study were: (a) to evaluate reliability of landmark location in digital processed images of film cephalograms of varying quality with that obtained in the films, and (b) to evaluate the pattern of error in landmark location with the two techniques. The sample consisted of 20 original cephalograms (OQ), 10 'high' quality images (Q1), and 10 'average' quality images (Q2). Measurements on films were performed using a digitizer connected to a PC and controlled by a cephalometric program, while measurements on digital images were obtained using a mouse and a cephalometric program for landmark sampling. To assess the digital images (DQ), the films were video recorded (B/W video camera), digitized (matrix = 512 x 512 x 8), stored in a PC, and displayed on a quality monitor. Fifteen cephalometric landmarks were recorded by three observers twice on each of the 20 images by the two methods. Mean x- and y-values were calculated for each landmark to obtain the best estimate of each landmark position. Comparisons between the OQ and DQ group, and between the subgroups of different film quality were performed (Wilcoxon's rank sum test). In general, measurements from the OQ images were more reliable than from the DQ images (P < 0.02). The pattern of recording error also differed between the two techniques, OQ images being more reliable along the x-axis (P < 0.01), while along the y-axis, reliability was not significantly different between the two groups. Q1 and Q2 images did not differ significantly in either group (P > 0.05). A difference was, however, observed between OQ2 and DQ2 images (P < 0.02), while this was not the case for OQ1 and DQ1 images. The generally lower reliability for measurements on the digital images could be ascribed to the much larger error, especially in the y-axis, for the Q2 images.