亚实性肺结节CT三维质量分析：观察者内及观察者间差异

BACKGROUND AND OBJECTIVE: Subsolid pulmonary nodules tend to exhibit considerably slower growth rates than solid lesions, nevertheless, higher malignancy probability. The diagnosis of indeterminate nodules largely depends on the growth evaluation during follow-up. The growth can manifested as an increase in size or the appearance and/or subsequent increase of solid components. The mass reflect the product of volume and density and can be more sensitive in growth evaluation. However, the repeatability needs a further validation. The purpose of this study is to assess the intra and inter-observer variability of mass measurement for subsolid nodules. METHODS: 80 subsolid nodules in 44 patients were retrospectively enrolled. Both the volume and mass were measured by two radiologists using blind method independently. Intra-observer and inter-observer variability were analyzed and compared by Bland-Altman method intra-class correlation test and Wilcoxon test. RESULTS: Software achieved satisfied segmentation for 92.5% nodules. Of them, 35% underwent manual modification. The 95% limits of agreement for intra-observer variability were -11.5%-10.4% for mass and -8.4%-8.8% for volume. The 95% limits of agreement for inter-observer variability were -17.4%-19.3% for mass and -17.9%-19.4% for volume.The intra-class correlation foefficients between volume and mass measument was 0.95 and 0.93 (both P < 0.001) and no significant differences (P=0.78, 0.09) was found for intra- and inter-observer variability. Manual modification of the segmentation caused the worse mass measurement repeatability in spite of the reader satisfaction. CONCLUSION: The repeatability of mass measurement has no significant difference with that of volume measurement and may act as a reliable method in the follow-up of subsolid nodules.