Cone Beam CT Measures Abdominal Tumor Motion Better Than 4DCT and Equal to On-table Fluoroscopy

Abstract

Although 4DCT is most often used for motion management, research shows it does not adequately predict on-table tumor motion for abdominal radiation therapy. In this study, we demonstrated the use of Cone Beam CT (CBCT) projection data for predicting intrafraction tumor motion and compared our results to pre and post treatment fluoroscopy (FL). For 31 patients with abdominal tumors and implanted fiducial markers, tumor motion was measured with CBCT and fluoroscopy for 202 treatment fractions and compared to the planned tumor motion from 4DCT. We processed the CBCT projections using an in-house fiducial tracking algorithm and used maximum-likelihood to calculate the fiducial trajectories. The trajectories were also measured by pre and post treatment orthogonal fluoroscopy. The daily required internal margin (IM) was calculated for CBCT and fluoroscopy as the fifth through 95th percentiles of motion in each left-right (LR), superior-inferior (SI) and anterior-posterior (AP) direction. The planning IM4DCT (the range of fiducial motion on 4DCT) was determined to be adequate when it was within ±1.2 mm (AP, LR) and ±3 mm (SI) of the daily IMCBCT, IMpre-fluoro or IMfluoro (combined pre and post treatment fluoroscopy). To validate the CBCT technique as a predictive measure of intrafraction motion, we compared IMCBCT to IMpre-fluoro, which is our institutional standard for measuring daily intrafractional motion. Finally, we compared IMCBCT to IMfluoro to identify patients who could benefit from imaging during treatment. The planning 4DCT could not accurately predict the intrafractional tumor motion observed during CBCT in ≥80% of fractions for 97% of patients. As measured by IMpre-fluoro and IMfluoro, 4DCT failed to predict intrafractional tumor motion in ≥80% of fractions for 94% and 100% of patients, respectively. Comparing CBCT to fluoroscopy, IMCBCT was in agreement with or larger than IMpre-fluoro for 96.0%, 91.3% and 93.4% of fractions in LR, SI and AP directions, respectively. For conventionally fractionated patients (median treatment time, t = 8.8 min), IMCBCT was in agreement with or larger than IMfluoro for 97.3% of fractions (in SI direction); this decreased to 82.3% for SBRT patients (t = 17 min) demonstrating the need for imaging during treatment for patients with longer treatment times. Tumor motion determined from 4DCT-simulation does not accurately predict the daily motion observed on CBCT or fluoroscopy. Fiducial motion derived from CBCT projection data accurately reflects the motion observed on pre and posttreatment fluoroscopy; CBCT could replace fluoroscopy for pretreatment internal margin verification, potentially reducing patient setup time and imaging dose. For patients with longer treatment times, there is a need for imaging during treatment to verify internal margin accuracy.