Geometrical differences in target volumes between slow CT and 4D CT imaging in stereotactic body radiotherapy for lung tumors in the upper and middle lobe.

Since stereotactic body radiotherapy (SBRT) was started for patients with lung tumor in 1998 in our institution, x-ray fluoroscopic examination and slow computed tomography (CT) scan with a rotation time of 4 s have been routinely applied to determine target volumes. When lung tumor motion observed with x-ray fluoroscopy is larger than 8 mm, diaphragm control (DC) is used to reduce tumor motion during respiration. After the installation of a four-dimensional (4D) CT scanner in 2006, 4D CT images have been supplementarily acquired to determine target volumes. It was found that target volumes based on slow CT images were substantially different from those on 4D CT images, even for patients with lung tumor motion no larger than 8 mm. Although slow CT scan might be expected to fare well for lung tumors with motion range of 8 mm or less, the potential limitations of slow CT scan are unknown. The purpose of this study was to evaluate the geometrical differences in target volumes between slow CT and 4D CT imaging for lung tumors with motion range no larger than 8 mm in the upper and middle lobe. Of the patients who underwent SBR between October 2006 and April 2008, 32 patients who had lung tumor with motion range no larger than 8 mm and did not need to use DC were enrolled in this study. Slow CT and 4D CT images were acquired under free breathing for each patient. Target volumes were manually delineated on slow CT images (TV(slow CT)). Gross tumor volumes were also delineated on each of the 4D CT volumes and their union (TV(4D CT)) was constructed. Volumetric and statistical analyses were performed for each patient. The mean +/- standard deviation (S.D.) of TV(slow CT)/TV(4D CT) was 0.75 +/- 0.17 (range, 0.38-1.10). The difference between sizes of TV(slow CT) and TV(4D CT) was not statistically significant (P = 0.096). A mean of 8% volume of TV(slow CT) was not encompassed in TV(4D CT) (mean +/- S.D. = 0.92 +/- 0.07). The patients were separated into two groups to test whether the quality of target delineation on slow CT scans depends on respiratory periods below or above the CT rotation time of 4 s. No significant difference was observed between these groups (P = 0.229). Even lung tumors with motion range no larger than 8 mm might not be accurately depicted on slow CT images. When only a single slow CT scan was used for lung tumors with motion range of 8 mm or less, 95% confidence values for additional margins for TV(slow CT) to encompass TV(4D CT) were 4.0, 5.4, 4.9, 5.1, 1.8, and 1.7 mm for lateral, medial, ventral, dorsal, cranial, and caudal directions, respectively.