Purpose  

The aim of this study was to determine whether postprocessing techniques could improve the accuracy of detecting lung nodules.

Materials and methods  

A total of 154 segmented lung volumes of multidetector-row computed tomography (MDCT) data were the subject of the study. Lung nodules were present in 88 volumes and absent in 66 volumes. We prepared four groups: (1) 7- or 10-mm thick-section axial images; (2) 1-mm thin-section axial images; (3) sliding slab maximum intensity projection (MIP) images with a slab thickness of 15 mm; and (4) sliding slab volume rendering (VR) images with a slab thickness of 15 mm. Sixteen physicians reviewed each group in interactive cine mode. The observers’ performance in the detection of lung nodule was evaluated by receiver operating characteristic (ROC) analysis.

Results  

The observers’ performance of the MIP and VR groups was significantly better than in other two groups. There was no significant difference statistically between the thin and thick groups.

Conclusion  

The detectability of lung nodules is improved with the use of sliding slab MIP and VR using thin-section image data. Thin-section volume data are essential for improving diagnostic accuracy, but observation of thin-section images without utilization of image-processing techniques dose not improve diagnostic accuracy.