2010 年 30 巻 1 号 p. 3-9
We propose a new method of converting thin-slice CT images to thick-slice images without image raw data. Slice sensitivity profiles (SSPs) were measured in the thin- and thick-slice settings. The spatial frequency property in the slice direction of the thin-slice images was converted to that of thick-slice images, based on the spatial frequency response calculated using the each SSP. In this study, we used CT images of seven subjects obtained in lung cancer screening, with a 4-slice scanner. The images were reconstructed with slice thicknesses of 5 and 8 mm, and the SSPs were measured in the scanner for the two slice settings, respectively. We calculated 8 mm-thick images from the 5 mm-thick images by the proposed method, and compared them with the actual 8 mm-thick images; this difference was evaluated by the root mean square error (RMSE). The mean value of RMSEs for seven subjects was approximately 6.0 Hounsfield Unit (HU). In addition, we obtained 8 mm-thick images by a conventional technique of weighted average from the 5 mm-thick images; the mean value of RMSEs was approximately 10.0 HU. The proposed method was demonstrated to be effective for converting thin-slice images to thick-slice images accurately.