Clinical comparison of adaptive 4DCBCT scanning protocols for lung tumor motion assessment

Adaptive four-dimensional cone beam computed tomography (4DCBCT) has been proposed as a novel method to reduce imaging dose and scan time. This technology involves both adaptive imaging and motion-compensated reconstruction. However, no study has been performed in a lung cancer patient cohort to confirm that adaptive 4DCBCT accurately images tumor motion. This is partly due to a lack of a ground truth comparison and the difficulty of assessing 4DCBCT images. This issue will be addressed in this study.

This study aims to investigate and assess lung tumor motion in adaptive four-dimensional cone beam computed tomography (4DCBCT) across two treatment days using data from the ADAPT clinical trial. Tumor motion measured in reconstructed adaptive 4DCBCT images was compared to motion observed in conventional 4-min 4DCBCT scans to evaluate the accuracy and feasibility of adaptive imaging for clinical use.

Lung tumor motion from the ADAPT clinical trial was compared between conventional 4DCBCT (1320 projections in 4 min) and adaptive 4DCBCT for different adaptive scan types [a fast 200 projection adaptive 4DCBCT scan (ADAPT_200) acquired in approximately 60 s and a slower, 600 projections adaptive 4DCBCT scan acquired in approximately 4 min (ADAPT_600)]. The fast 200 projection adaptive 4DCBCT was reconstructed using a motion-compensated image reconstruction algorithm, while the conventional and slower 600 projection scans were reconstructed with the FDK algorithm. Analysis was performed across two fractions (f1 and f2) on different days. For each scan type, the center of mass (COM) trajectory was computed for each respiratory phase to quantify tumor motion. The mean and standard deviation difference in tumor COM between the conventional and adaptive 4DCBCT scans were used to quantify tumor trajectory accuracy.

For the fast adapt scan (ADAPT_200), 80.4% of fractions showed motion differences of less than 1 mm from the conventional scan, 13.3% had motion differences of 1-2 mm, and 6.7% had motion differences greater than 2 mm. For the high-quality adapt scan (ADAPT_600), 76.9% of fractions showed motion differences of less than 1 mm from the conventional scan, 15.4% had motion differences of 1-2 mm, and 7.7% had motion differences greater than 2 mm.

Adaptive 4DCBCT with as low as 200 projections can image the tumor's trajectory, however, minor deviations were observed, with 13.3% of cases showing motion differences of 1-2 mm and 6.7% exceeding 2 mm. These findings demonstrate that faster acquisition (60 s) of adaptive 4DCBCT is feasible for a majority of lung cancer radiation therapy patients.