Management of respiration-induced motion with 4-dimensional computed tomography (4DCT) for pancreas irradiation

The purposes of this study were to quantify respiration-induced organ motions for pancreatic cancer patients and to explore strategies to account for these motions.

Both 3-dimensional computed tomography (3DCT) and 4-dimensional computed tomography (4DCT) scans were acquired sequentially for 15 pancreatic cancer patients, including 10 randomly selected patients and 5 patients selected from a subgroup of patients with large tumor respiratory motions. 3DCTs were fused with 2 sets of 4DCT data at the end of exhale phase (50%) and the end of inhale phase (0%). The target was delineated on the 50% and 0% phase CT sets, and the organs at risk were drawn on the 3DCT. These contours were populated to the CT sets at other respiratory phases based on deformable image registration. Internal target volumes (ITV) were generated by tracing the target contours of all phases (ITV10), 3 phases of 0%, 20% and 50% (ITV3), and 2 phases of 0% and 50% (ITV2). ITVs generated from phase images were compared using percentage of volume overlap, Dice coefficient, geometric centers, and average surface distance.

Volume variations of pancreas, kidneys, and liver as a function of respiratory phases were small (<5%) during respiration. For the 10 randomly selected patients, peak-to-peak amplitudes of liver, left kidney, right kidney, and the target along the superior-inferior (SI) direction were 7.9 ± 3.2 mm, 7.1 ± 3.1 mm, 5.7 ± 3.2 mm, and 5.9 ± 2.8 mm, respectively. The percentage of volume overlap and Dice coefficient were 92% ± 1% and 96% ± 1% between ITV10 and ITV2 and 96% ± 1% and 98% ± 1% between ITV10 and ITV3, respectively. The percentage of volume overlap between ITV10 and ITV3 was 93.6 ± 1.1 for patients with tumor motion >8 mm.

Appropriate motion management strategies are proposed for radiation treatment planning of pancreatic tumors based on magnitudes of tumor respiratory motions.