Pulmonary nodules: Preliminary experience with three-dimensional evaluation

PURPOSE: To evaluate software designed to calculate pulmonary nodule volume in three dimensions. MATERIALS AND METHODS: Fifty-four solid noncalcified pulmonary nodules measuring 5-18 mm in diameter were studied with computed tomographic (CT) volumetric software. Baseline CT examinations were performed for various indications by using four-detector row multisection CT units, 1.25- or 2.50-mm sections, and a standard reconstruction algorithm. The percentage of successful nodule segmentations, as well as intraobserver variability, interreader agreement, and global repeatability of calculated volumes, was determined on the basis of consecutive measurements performed three times by three different radiologists by using the Bland and Altman method. The software was used to calculate the doubling time of 22 nodules for which a final diagnosis and comparable CT scans were available. RESULTS: Fifty-two (96%) of the 54 nodules were successfully segmented, allowing their volume to be calculated. Repeatability was high: There was no variation in the nine measurements of 35 (67%) of the 52 nodules. The coefficient of variation for the remaining 17 nodules (33%) was 2.26%. Bland and Altman 95% limits of acceptability, calculated on the basis of log-transformed data, yielded a maximum software measurement error of 6.38% of the previous volume measurement. Doubling time ranged from 4 to 188 years for the 13 benign nodules and from 37 to 216 days for the nine malignant nodules. CONCLUSION: Software volumetric analysis yielded repeatable estimates for 96% of the nodules examined. All software-calculated doubling times were in keeping with the benign or malignant nature of the nodules. (C) RSNA, 2004.