Fig 1.
Flowchart of the study population.
Fig 2.
Semi-automated measurement of pulmonary nodules.
(A) Baseline low dose chest CT (LDCT, right side) and follow-up LDCT (left side) show a solid nodule at right lower lobe of the lung. (B) Semi-automated segmentation and measurement of the nodule can be initiated by drawing a line on the nodule. (C) The software can visualize the result of semi-automated segmentation as well as the result of measurement. On the baseline LDCT, average diameter and volume of the nodule were 4.6 mm and 52.0 mm3, respectively. (D) On the follow-up LDCT, the average diameter and volume of the nodule were 5.3 mm and 76.1 mm3, respectively. Consequently, absolute diameter growth and percentage volume growth of the nodule were 0.7 mm and 31.7% respectively, and the volume doubling time was 602 days.
Table 1.
Volume and average diameter of the 115 nodules measured in baseline and follow-up screening low-dose chest CT.
Fig 3.
(A) Receiver operating characteristic (ROC) curves of the volumetric and diametric measurement for diagnosing lung cancers. The area under the curve (AUC) values of growth adjudicated by volumetric and diametric measurement for diagnosing lung cancers were 0.812 and 0.810, respectively (p = 0.995). (B) ROC curve of volume doubling time for diagnosing lung cancer (AUC, 0.793). The sensitivity and specificity of the radiologist’s diagnostic referral were 76.9% and 96.6%, respectively.
Table 2.
Comparison of diagnostic performance for lung cancer diagnosis between growths adjudication of volumetric and diametric measurements, and subjective radiologist’s assessment.
Table 3.
Diagnostic performance of volume doubling time for lung cancer diagnosis in 115 indeterminate nodules detected in baseline screening CT.
Fig 4.
Bland-Altman plots for agreement (A) between mean diameters measured by two radiologists and difference of diameters between the two radiologists (the 95% limit of agreement was between -2.9mm to 2.6mm), and (B) between mean volumes measured by two radiologists and difference of volumes between the two radiologists (the 95% limit of agreement was between -85.5mm3 and 107.4mm3).