PD-0231: Time-resolved internal target translation and rotation during liver SBRT treatments

J. Bertholet, W. Fledelius, E.S. Worm, M. Høyer, P.R. Poulsen
2015 Radiotherapy and Oncology  
S116 3rd ESTRO Forum 2015 the the dominant Cranio-Caudal(CC) direction of motion were optimized for 4 NSCLC lung cancer patients in Pinnacle v9.1. The plans were made on mid-position CT with iGTV density override, spot size σ =6 mm and spot spacing 1σ. The total prescribed dose was 66Gy in 24 fractions to be delivered to >95% of planning target volume (PTV). The fractionated proton treatment delivery was simulated by an in-house developed time dependent simulation algorithm which integrates 4D
more » ... eformation vector field (DVF) and realistic respiratory trace determined from CBCT images. The DVF describes the anatomical changes of each point over the respiratory cycle as a cyclic function. Each cycle is synchronously modified with the phase, scale and offset information from the real respiratory trace. The impact of the interplay effect in our simulation was reproduced through Dose with Interplay to be compared with a simple shift invariant blurring of the dose the Blurred Dose. We used the following formulas for quantifying the impact of interplay and range (blurring) effects on dose errors: Results: Dose errors introduced by the interplay effect demonstrated different relationship with the respiration amplitude compared to the blurring effects. The errors due to interplay effects showed linear relationship with the amplitude in all patients. The blurring effect on the other side caused dose error with quadratic dependence on the amplitude (Figure 1 ). Our preliminary data also suggest that the dose errors due to interplay effects exhibit faster rise rate with the amplitude as the tumor size increases. Conclusions: The dosimetric consequences from the interplay and range effects in proton therapy are related to the amplitude. Large tumors with higher amplitude of motion seem to be more sensitive to dose errors. Purpose/Objective: A high accuracy is crucial in stereotactic body radiotherapy (SBRT) treatments, but may be compromised by both translations and rotations of the target during treatment. While target translations have been investigated in detail, only few studies have addressed rotations. This study presents the first measurements of time-resolved intra-fraction internal target translations and rotations during liver SBRT using CBCT projections. Materials and Methods: Twenty-seven patients with three implanted gold markers received three to six-fraction SBRT on a conventional LINAC. One to three CBCT scans were acquired during each fraction (186 CBCTs in total). The three markers were retrospectively automatically segmented on the CBCT projections using in-house developed software. The 3D trajectory of each marker was estimated using a probability based method. Intra-fraction translation and rotation of the marker constellation with respect to the mean position over the whole scan were calculated using Singular Value Decomposition (SVD) (Fig 1.) . Motion range and Pearson's correlation coefficients (R) per scan between the six degrees of freedom were investigated for patients with motion amplitude exceeding 1mm. Results: Figure 1 shows an example of the time-resolved translation and rotation of the marker constellation during a CBCT acquisition. Table 1 presents the 2-98 percentile range of translation along and rotation around the Right-Left (RL), Superior-Inferior (SI) and Anterio-Posterior (AP) axis. The motion was the largest in the SI direction (mean range of 8.93 mm) while the rotation was of similar magnitude around all three axes. Rotations higher than 10 degrees were occasionally observed. The highest correlation was observed between AP and SI translations (Fig 1.) with mean R = -0.85 (SD = 0.22). The negative correlation means that cranial motion is associated with posterior motion. There was significant correlation (either positive or negative) between the SI translation and RL rotation in general, except for five patients with substantial cardiac induced motion for at least one marker (gray bars in figure c). Thus cranial motion is associated with rotation around the RL axis in either direction, but consistently the same direction for the same patient. Conclusions: Highly time resolved translations and rotations of targets in liver SBRT were determined for the first time on a conventional LINAC using CBCT projections. Considerable intra-fraction translations and rotations were observed.
doi:10.1016/s0167-8140(15)40229-4 fatcat:d5xflv62yfdv7fib7atvk4jaem