Influence of contrast enhancement and pseudocolor transformation on the diagnosis with digital subtraction images (DSI).

Abstract

This study evaluated the influence of the image processing step of digital subtraction images on inter- and intraexaminer agreement in the interpretation of alveolar bone changes. 52 pairs of standardized radiographs from various clinical trials were included. Six dentists were invited to interpret the images projected as slides in random order. Display one demonstrated the slide of the digitized baseline radiograph and the follow-up image. Display two showed the regular digital subtraction image. Display three represented a grey level contrast enhanced version and displays four and five were pseudo/color enhanced subtraction images. Applying kappa-statistics and multiple regression analysis it was demonstrated that better agreements were obtained when the two color coded displays of subtraction images were shown to the interpreters. The image interpretation was performed in two series. For the first evaluation the interpreters were not informed about the therapy provided nor the time elapsed between taking the pairs of standardized radiographs. In the second series this information was provided. It could be demonstrated that the agreement in the diagnosis of bone change was less influenced by the knowledge about the clinical information if the two color-converted versions of subtracted images were evaluated. Thus, it was concluded that image processing of subtraction images using color enhancement might improve agreement in the diagnostic task. The color coded images were less influenced by the bias in the interpretation of an expected change. The decision making process might be more objective when using color enhanced subtraction images.