Effects of Irregular Respiratory Motion on the Positioning Accuracy of Moving Target with Free Breathing Cone-Beam Computerized Tomography

Abstract

For positioning a moving target, a maximum intensity projection (MIP) or average intensity projection (AIP) image derived from 4DCT is often used as the reference image which is matched to free breathing cone-beam CT (FBCBCT) before treatment. This method can be highly accurate if the respiratory motion of the patient is stable. However, a patient’s breathing pattern is often irregular. The purpose of this study is to investigate the effects of irregular respiration on positioning accuracy for a moving target aligned with FBCBCT. Nine patients’ respiratory motion curves were selected to drive a Quasar motion phantom with one embedded cubic and two spherical targets. A 4DCT of the phantom was acquired on a CT scanner (Philips Brilliance 16) equipped with a Varian RPM system. The phase binned 4DCT images and the corresponding MIP and AIP images were transferred into Eclipse for analysis. FBCBCTs of the phantom driven by the same respiratory curves were also acquired on a Varian TrueBeam and fused such that both CBCT and MIP/AIP images share the same target zero positions. The sphere and cube volumes and centroid differences (alignment error) determined by MIP, AIP and FBCBCT images were calculated, respectively. Compared to the volume determined by MIP, the volumes of the cube, large sphere, and small sphere in AIP and FBCBCT images were smaller. The alignment errors for the cube, large sphere and small sphere with center to center matches between MIP and FBCBCT were 2.5 ± 1.8mm, 2.4±2.1 mm, and 3.8±2.8 mm, and the alignment errors between AIP and FBCBCT were 0.5±1.1mm, 0.3±0.8mm, and 1.8±2.0 mm, respectively. AIP images appear to be superior reference images to MIP images. However, irregular respiratory pattern could compromise the positioning accuracy, especially for smaller targets.

A modern review of the uncertainties in volumetric imaging of respiratory‐induced target motion in lung radiotherapy

“…On the contrary, a different phantom study performed by Shirai et al, 226 recommended using the AIPbased ITV because of larger misalignment errors found for MIP-ITV vs AIP-ITV matching to the CBCT. Li et al 227 also confirmed these results and found AIP-ITV matching as the superior reference, based on their phantom study comparing MIP and AIP with 9 patient's respiratory profiles. Cai et al 228 reported the same results of significant differences between MIP and CBCT-based ITVs using a digital human phantom for irregular breathing cycles across all different tumor sizes.…”

Dynamic cone-beam CT reconstruction using spatial and temporal implicit neural representation learning (STINR)

“…Such irregularity may lead to substantial intra-phase motion variations and strong residual motion artifacts after sorting (Cooper et al 2015 ). The nominal cycle resolved by 4D-CBCT fails to capture the irregularity and non-periodicity, which may provide crucial information on motion statistics and trends to guide patient immobilization, set-up, and treatment monitoring (Poulsen et al 2014 , Li et al 2018 ). The ultimate solution to such a challenge is time-resolved CBCT imaging, or dynamic CBCT (Li et al 2010 , Cai et al 2014 , Gao et al 2018 , Jailin et al 2021 ).…”

Dynamic CBCT imaging using prior model-free spatiotemporal implicit neural representation (PMF-STINR)

“…The motion sorting assumes that the underlying anatomical motion is periodic and regular, which is in general false (Yasue et al 2022 ). Correspondingly, 4D-CBCT cannot capture time-resolved irregular motion which may significantly impact patient setup and dose delivery accuracy (Vergalasova et al 2011 , Clements et al 2013 , Li et al 2018 ). Moreover, 4D motion sorting is usually based on surrogates (e.g.…”