Sources of noise in digital subtraction radiography.

In 17 patients, three identical radiographic examinations were performed for the lower premolar-molar region, by two examiners at one visit and by one examiner again approximately 3 months later. Recordings were performed with a newly developed device. The radiographs were converted to digital images in the following way: (1) video camera recording of the radiograph obtained at first visit by the first examiner (A), (2) copying of the numeric image of this radiograph (B), (3) repeated video recording of this radiograph (C), (4) video recording of the radiograph obtained at first visit by the second examiner (D), and (5) video recording of the radiograph obtained after 3 months (E). Subtractions were performed between the images: B - A, C - A, D - A, and E - A. The standard deviation of the histogram for the distribution of gray shades in the subtraction images was used to evaluate noise in the image. The standard deviation was 1.3 between B - A, 2.8 between C - A, 4.0 between D - A, and 4.1 between E - A. Physical noise originating from the video camera and the analog-to-digital conversion process thus constituted the major part of noise seen in subtraction images performed between radiographs recorded with a time interval. This indicates that reproducible recordings could be performed with the present device, which, in combination with a subtraction program able to correct for some geometry and density differences, may aid the use of subtraction radiography in clinical trials.