Shinichi Shimizu, Hiroki Shirato, Kenji Kagei, Takeshi Nishioka, Xo Bo, Hirotoshi Dosaka-Akita, Seiko Hashimoto, Hidefumi Aoyama, Kazuhiko Tsuchiya, Kazuo Miyasaka

Three-dimensional (3D) treatment planning has often been performed while patients breathe freely, under the assumption that the computed tomography (CT) images represent the average position of the tumor. We investigated the impact of respiratory movement on the free-breathing CT images of small lung tumors using sequential CT scanning at the same table position. Using a preparatory free-breathing CT scan, the patient's couch was fixed at the position where each tumor showed its maximum diameter on image. For 16 tumors, over 20 sequential CT images were taken every 2 s, with a 1-s acquisition time occurring during free breathing. For each tumor, the distance between the surface of the CT table and the posterior border of the tumor was measured to determine whether the edge of the tumor was sufficiently included in the planning target volume (PTV) during normal breathing. In the sequential CT scanning, the tumor itself was not visible in the examination slice in 21% (75/357) of cases. There were statistically significant differences between lower lobe tumors (39.4%, 71/180) and upper lobe tumors (0%, 0/89) (p < 0.01) and between lower lobe tumors and middle lobe tumor (8.9%, 4/45) (p < 0.01) in the incidence of the disappearance of the tumor from the image. The mean difference between the maximum and minimum distances between the surface of the CT table and the posterior border of the tumor was 6.4 mm (range 2.1-24.4). Three-dimensional treatment planning for lung carcinoma would significantly underdose many lesions, especially those in the lower lobe. The excess "safety margin" might call into question any additional benefit of 3D treatment. More work is required to determine how to control respiratory movement.