The aim of this study was to evaluate thoracic computed tomography (CT) images compressed by different joint photographic experts group (JPEG) algorithms at ratios of 10:1, 20:1, 30:1, and 40:1. As a physical evaluation, the modulation transfer function (MTF) of these images was calculated, and, as a clinical evaluation, 24 chest CT images (12 control subjects and 12 with pulmonary nodules) were used. Five radiologists interpreted the presence or absence of nodules and evaluated the quality of images. The detection of nodules was studied by receiver operating characteristic (ROC) and power spectral analyses. MTF was decreased at compression ratios higher than 20:1, owing to the influence of blocking artifacts. There was no significant difference in nodule detection by ROC and power spectral analysis between uncompressed and compressed CT images. The quality of images compressed at ratios higher than 20:1 was inferior. JPEG algorithms allow the compression of CT images without compromising their diagnostic value at a ratio of 20:1.
