[Evaluation of data compression in gastrointestinal examinations using computed radiography].

The purpose of this study was to evaluate the clinical utility of Type III irreversible data compression of computed radiography (CR) in gastrointestinal examinations. Simulated gastric lesions created on a specially made phantom gastric mucosa as well as normal and abnormal cases selected from upper gastrointestinal radiographic examinations and barium enemas were used. Twenty-five sets of original and compressed CR images were obtained for each category of normal and abnormal cases. Visual ranking and a receiver operating characteristic (ROC) study using these images were carried out by 5 radiologists. By visual ranking, deterioration of image quality on the compressed images was recognized in 32.8 to 58.4% of the phantom gastric mucosa images, 23.2 to 75.2% of the upper gastrointestinal examinations, and 21.6 to 81.6% of the barium enemas. However, this deterioration was subtle and the analysis of ROC curves revealed no significant difference in detectability of the lesion between original and compressed images. It is concluded that this Type III irreversible data compression is clinically applicable and useful in gastrointestinal examinations using CR.