Variability of pulmonary nodule volumetry on coronary CT angiograms

Figures

• 

  Figure 1.. The importance of measurement variability in growth estimation. All graphs present expected volume measurements of 3 pulmonary nodules with different growth-rates [i.e., VDT = 30 days, as in inflammatory changes (green); VDT = 250 days, as in malignancy (red); and VDT = 600 days, as in benign pathology (blue)]. The shaded areas represent possible volume measurements over time given the different measurement variability values [left (top and bottom)—25%; top right—10%, and bottom right—50%]. Volume measurements in overlapping shaded areas cannot be confidently attributed to 1 growth rate or another. The starting volume of a nodule [top left—80 cc3; bottom left—150 cc3] does not change the optimal waiting time between follow-up scans for a confident discrimination between suspicious and benign pathology (dashed orange line), but the measurement variability can significantly shorten (top right) or extend it (bottom right).

• 

  Figure 2.. Example of a nodule included in the study measured in systole (A) and diastole (B). The seed points (*) were placed in the nodule center, and the yellow lines show the semiautomatic segmentation result, which was not corrected and considered adequate by both observers. Notice the small vessel (arrows) approaching the nodule, which was excluded from the segmentation in both instances. The volume measured in diastole was 80% greater than in systole.

• 

  Figure 3.. Relative (percent) Bland-Altman plots, with estimated limits of agreement, when comparing volume measurements between systole and diastole (top left corner), different software packages (top right corner), and different observers (bottom). Interobserver subanalysis is presented for the systolic (bottom left corner) and diastolic datasets separately (bottom right corner).