## Abstract

Purpose/Objective(s):

In motion inclusive radiotherapy, respiratory tumor motion is normally accounted for either by population-based CTV-PTV margins that are parallel to the room coordinate system, or by introduction of an ITV that encompasses the tumor motion in a planning CT scan. While the former ignores motion correlation along different axes the latter tends to overestimate the dosimetric consequences of random motion. The purpose of this study was to propose and investigate an individualized adaptive margin approach that accounts for motion correlation while still considering the different impacts of random and systemic motion.

Materials/Methods:

The study used a tumor motion database with 160 abdominal/thoracic tumor trajectories (46 patients) acquired with the Cyberknife Synchrony system. The following treatment scenario was simulated for the 138 trajectories that exceeded 15 minutes: First, a cone-beam CT (CBCT) scan with 600 projections was acquired in 1 minute. The tumor position in the CBCT projections was used to estimate the mean position and covariance matrix of the tumor motion (i.e., tumor motion magnitude and tumor motion axes) during the CBCT acquisition. Next, the patient was aligned to the estimated mean tumor position. A PTV was constructed from a 2 cm spherical CTV by adding margins parallel to the estimated tumor motion axes. The margins were calculated as 0.7σ + 2.5∑. Here, ∑ is the SD of the population-based systematic motion-induced errors along the tumor motion axes, while σ is the SD of random motion-induced errors along the tumor motion axes. The applied value for σ was the largest of the population-based random motion and the individually CBCT estimated random motion. The individual value for σ was largest (and therefore used for margin calculation) in 40% of the cases. For each trajectory, the accumulated CTV surface dose distribution and minimum CTV dose were calculated for the time interval form 1 to 15 min (10 Hz time resolution). For comparison, the same quantities were calculated for standard population-based margins parallel to the room coordinate system.

Results:

The minimum accumulated CTV dose Dmin was >95% in 86% of the cases, >94% in 92% of the cases, and >90% in 98% of the cases. The smallest Dmin was 84%. For standard margins, Dmin was >95% in 82% of the cases, >94% in 84% of the cases, and >90% in 93% of the cases. Here, the smallest Dmin was 71%. The mean PTV volume was 7.6 cm3 for adaptive margins (4.3 mm, 1.6 mm, 1.0 mm margins) and 8.2 cm3 for standard margins (1.8 mm, 3.4 mm, 2.4 mm margins).

Conclusions:

A strategy for individualized adaptive margins that accounts for motion correlation was proposed. Compared to standard margins, the adaptive margins result in slightly smaller PTvs. and give fewer outliers with low minimum CTV dose.

In motion inclusive radiotherapy, respiratory tumor motion is normally accounted for either by population-based CTV-PTV margins that are parallel to the room coordinate system, or by introduction of an ITV that encompasses the tumor motion in a planning CT scan. While the former ignores motion correlation along different axes the latter tends to overestimate the dosimetric consequences of random motion. The purpose of this study was to propose and investigate an individualized adaptive margin approach that accounts for motion correlation while still considering the different impacts of random and systemic motion.

Materials/Methods:

The study used a tumor motion database with 160 abdominal/thoracic tumor trajectories (46 patients) acquired with the Cyberknife Synchrony system. The following treatment scenario was simulated for the 138 trajectories that exceeded 15 minutes: First, a cone-beam CT (CBCT) scan with 600 projections was acquired in 1 minute. The tumor position in the CBCT projections was used to estimate the mean position and covariance matrix of the tumor motion (i.e., tumor motion magnitude and tumor motion axes) during the CBCT acquisition. Next, the patient was aligned to the estimated mean tumor position. A PTV was constructed from a 2 cm spherical CTV by adding margins parallel to the estimated tumor motion axes. The margins were calculated as 0.7σ + 2.5∑. Here, ∑ is the SD of the population-based systematic motion-induced errors along the tumor motion axes, while σ is the SD of random motion-induced errors along the tumor motion axes. The applied value for σ was the largest of the population-based random motion and the individually CBCT estimated random motion. The individual value for σ was largest (and therefore used for margin calculation) in 40% of the cases. For each trajectory, the accumulated CTV surface dose distribution and minimum CTV dose were calculated for the time interval form 1 to 15 min (10 Hz time resolution). For comparison, the same quantities were calculated for standard population-based margins parallel to the room coordinate system.

Results:

The minimum accumulated CTV dose Dmin was >95% in 86% of the cases, >94% in 92% of the cases, and >90% in 98% of the cases. The smallest Dmin was 84%. For standard margins, Dmin was >95% in 82% of the cases, >94% in 84% of the cases, and >90% in 93% of the cases. Here, the smallest Dmin was 71%. The mean PTV volume was 7.6 cm3 for adaptive margins (4.3 mm, 1.6 mm, 1.0 mm margins) and 8.2 cm3 for standard margins (1.8 mm, 3.4 mm, 2.4 mm margins).

Conclusions:

A strategy for individualized adaptive margins that accounts for motion correlation was proposed. Compared to standard margins, the adaptive margins result in slightly smaller PTvs. and give fewer outliers with low minimum CTV dose.

Originalsprog | Engelsk |
---|---|

Publikationsdato | 1 nov. 2010 |

Antal sider | 2 |

Status | Udgivet - 1 nov. 2010 |

Begivenhed | ASTRO 52nd Annual Meeting - San Diego, USA Varighed: 31 okt. 2010 → 4 nov. 2010 |

### Konference

Konference | ASTRO 52nd Annual Meeting |
---|---|

Land/Område | USA |

By | San Diego |

Periode | 31/10/2010 → 04/11/2010 |