Respiratory-Gated Helical Computed Tomography of Lung: Reproducibility of Small Volumes in an Ex Vivo Model

Abstract

Motion-adapted radiotherapy with gated irradiation or tracking of tumor positions requires dedicated imaging techniques such as four-dimensional (4D) helical computed tomography (CT) for patient selection and treatment planning. The objective was to evaluate the reproducibility of spatial information for small objects on respiratory-gated 4D helical CT using computer-assisted volumetry of lung nodules in a ventilated ex vivo system. Five porcine lungs were inflated inside a chest phantom and prepared with 55 artificial nodules (mean diameter, 8.4 mm +/- 1.8). The lungs were respirated by a flexible diaphragm and scanned with 40-row detector CT (collimation, 24 x 1.2 mm; pitch, 0.1; rotation time, 1 s; slice thickness, 1.5 mm; increment, 0.8 mm). The 4D-CT scans acquired during respiration (eight per minute) and reconstructed at 0-100% inspiration and equivalent static scans were scored for motion-related artifacts (0 or absent to 3 or relevant). The reproducibility of nodule volumetry (three readers) was assessed using the variation coefficient (VC). The mean volumes from the static and dynamic inspiratory scans were equal (364.9 and 360.8 mm3, respectively, p = 0.24). The static and dynamic end-expiratory volumes were slightly greater (371.9 and 369.7 mm3, respectively, p = 0.019). The VC for volumetry (static) was 3.1%, with no significant difference between 20 apical and 20 caudal nodules (2.6% and 3.5%, p = 0.25). In dynamic scans, the VC was greater (3.9%, p = 0.004; apical and caudal, 2.6% and 4.9%; p = 0.004), with a significant difference between static and dynamic in the 20 caudal nodules (3.5% and 4.9%, p = 0.015). This was consistent with greater motion-related artifacts and image noise at the diaphragm (p <0.05). The VC for interobserver variability was 0.6%. Residual motion-related artifacts had only minimal influence on volumetry of small solid lesions. This indicates a high reproducibility of spatial information for small objects in low pitch helical 4D-CT reconstructions.