Abstract
Background:
The detection of small nodules in pulmonary multidetector computed tomography (MDCT) data sets is challenging, and there is a need for visualization techniques that can improve reader sensitivity and efficiency. We have developed a subvolume rendering technique (“Softslice”) with nonlinear, symmetrical modulation of the relative signal intensity along the viewing direction. Our hypothesis was that this technique might provide an advantageous visual differentiation between nodules and overlapping vessels in pulmonary MDCT subvolumes.
Purpose:
To compare the Softslice subvolume rendering technique and maximum intensity projection (MIP) rendering for visualization of artificial pulmonary nodules in MDCT data sets.
Material and Methods:
Virtual, artificial pulmonary nodules were created with custom-made software. The nodules had the same signal intensity as pulmonary vessels. Approximately 50 nodules with diameter 2–7 mm were inserted in the Digital Imaging and Communications in Medicine (DICOM) image files of pulmonary MDCT data sets obtained without intravenous contrast. Each data set contained approximately 300 slices with 1-mm slice spacing. Four types of subvolumes were rendered: MIP with 10- and 20-mm subvolume thickness, and Softslice with 20- and 30-mm subvolume thickness. Three radiologists performed readings with free scrolling using 1-, 4- or 8-mm increments between subvolume positions.
Results:
No significant differences in the number of detected nodules were found between the different renderings with free scrolling using 1- and 4-mm increments. When 8-mm increment was used, nodule detection with the 30-mm Softslice rendering was significantly superior to the other renderings (P = 0.012 vs. MIP 10 mm, P = 0.018 vs. MIP 20 mm, P = 0.021 vs. Softslice 20 mm). When the subvolume increments were changed from 4 mm to 8 mm, the most marked reduction in the number of detected nodules was seen for MIP subvolumes with 10-mm thickness (P = 0.018).
Conclusion:
The Softslice rendering technique is promising for the detection of pulmonary nodules in MDCT data sets, and may allow more efficient reading than the standard MIP subvolume technique. With 10-mm MIP rendering, the detection rate for small pulmonary nodules substantially influenced by the incremental distance between subvolumes positions.