Irreversible data compression in chest imaging using computed radiography: an evaluation.

We studied the feasibility of a recently developed 20:1 to 25:1 compression technique for computed radiography (CR) using lung phantom images, and clinical cases were evaluated. We compared one original and two reconstructed CR films of the same image using compression ratios of 10:1 to 15:1, and 20:1 to 25:1, and the deterioration of image quality was carefully checked. Evaluation by subjective ranking of various abnormal opacities created for phantoms and clinical cases showed deterioration of image quality with 20:1 to 25:1 compression images. This was greatest for a single linear opacity (thin line) and least for reticulonodular opacities. Receiver operating characteristics (ROC) for pneumothorax, pulmonary nodules and reticulonodular opacities in newborns, infants, and adults revealed no significant differences among the original 10:1 to 15:1, and 20:1 to 25:1 compression images. We conclude that this new method of data compression of CR images is clinically applicable and useful.