The accuracy of extracted target motion trajectories in four-dimensional cone-beam computed tomography for lung cancer patients.

Abstract

To quantify the accuracy of extracted target motion trajectories in dual-source four-dimensional cone-beam computed tomography (4D-CBCT) by comparison with the actual three-dimensional (3D) target motion acquired simultaneously during 4D-CBCT scan. 4D-CBCT scans were performed for 19 different sinusoidal-like patterns and 13 lung cancer patients with implanted markers. Internal (In) or external (Ex) surrogates with amplitude (Amp)- or phase (Ph)-based sorting were used for the reconstructions. The targets were a pseudo-tumor and implanted marker for the phantom and clinical studies, respectively. The accuracy was evaluated by determining the maximum error (MaxE_i) between the 3D target position extracted from 4D-CBCT and the actual 3D target position detected by fluoroscopy in each ith phase (0⩽i⩽7). Median peak-to-peak target displacements in the superior-inferior (SI) direction were 20.6 and 20.6mm in the phantom and clinical studies, respectively. In the phantom and clinical studies, the maximum of median MaxE_is in the SI direction was 4.6 and 9.2mm in the In_Ph reconstruction. In the clinical study, the maximum of median MaxE_is was observed during the end-inhalation phase among all reconstruction approaches. This study showed the magnitude of underestimation toward the inferior direction of target motion in clinical 4D-CBCT.