An "in silico" clinical trial comparing free breathing, slow and respiration correlated computed tomography in lung cancer patients.

BACKGROUND AND PURPOSE To determine which method of internal target volume (ITV) definition based on a respiration correlated CT (RCCT) allows optimal tumor coverage. MATERIAL AND METHODS A free breathing CT (CT(fb)) and an RCCT scan were acquired in 41 lung cancer patients. For 12 patients with a motion >7 mm in any direction, a detailed analysis was made. The RCCT scan was used to measure tumor motion and to reconstruct a CT at 10 phases (CT(10ph)), amongst which the half ventilation CT (CT(hv)). By averaging the CT(10ph), a slow CT (CT(slow)) was reconstructed. Based on those scans ITVs were delineated and treatments were planned, where for the ITV(hv) an internal margin of (motion amplitude)/4 was used. The treatment plans for the ITVs were projected on the 10 respiration phases. Doses were calculated and averaged over the 10 phases to estimate the actual CTV coverage. RESULTS The 3D motion was on average 8.1+/-1.0 mm (1 SD) for all patients; no statistical difference was found between lower and upper lobe tumors. The ITV(slow) was the smallest volume on average (142+/-38 cm(3)), followed by the ITV(hv) (160+/-40 cm(3)), the ITV(10ph) (161+/-41 cm(3)) and the ITV(fb) (250+/-63 cm(3)). Mean CTV doses were between 95% and 107% of the prescribed dose for nearly all patients and treatment plans. Analysis of the CTV coverage suggested that underdosage may occur when the CT(slow) is used and a geographic miss occurred using the CT(fb), due to uncorrect localization of the average tumor position. CONCLUSIONS The CT(hv) seems to be the optimal dataset for delineation, using an adequate anisotropic internal margin of (motion amplitude)/4.