Effect of logarithmic contrast enhancement on subtraction images.

The aim of this study was to analyze logarithmic contrast enhancement as compared with regular digital subtraction and to consider effects of image texture and observer experience. Radiographs of three different textures, including computer-stimulated lesions of various sizes and densities, were subtracted from radiographs taken before lesions were placed. Logarithmically enhanced and regular subtraction images were put in random order with an equal number of nonlesion subtraction images to form a series of 300 images. Twenty subjects were asked to decide on a 5-point confidence scale whether a lesion was present. Plain texture images were diagnosed significantly better than bone and enamel images. Logarithmic contrast enhancement had a positive effect on plain texture images. There was no significant experience effect. In the student group, however, there was a significant difference between the two subtraction methods. Overall, logarithmic contrast enhancement provided better diagnostic validity than regular subtraction. Logarithmically enhanced subtraction radiography seems to be a promising method for images with low-contrast gradients.