Simulated dose reduction in conventional chest CT: validation study.

Abstract

To validate a technique of computer-simulated dose reduction for conventional chest computed tomography (CT). In 27 patients, CT scans were obtained at 200, 100, and 40 mAs at two levels. The raw data from the 200-mAs scan were modified on a computer workstation to simulate the increased noise present on 100- and 40-mAs scans. Real and simulated 100- and 40-mAs images were independently assessed in random order for overall image quality and radiologic findings by four subspecialty-trained chest radiologists who were blinded to the technique. The four observers were given paired real and simulated images. They were asked to identify the real image and note any difference in diagnostic quality. No difference was seen in overall image quality or radiologic findings between real and simulated images (P > .05). In the paired comparison, 433 of 864 (50.1%) real images were correctly identified. Computer modification of 200-mAs raw scan data to simulate 100- and 40-mAs noise levels produces reconstructed images indistinguishable from real 100- and 40-mAs scans. This technique provides realistic reduced-dose images without patient radiation exposure and with identical image registration and motion artifact.