Three-dimensional analysis of the respiratory interplay effect in helical tomotherapy: Baseline variations cause the greater part of dose inhomogeneities seen

Authors

  • Tudor G. Samuel J.,

    Corresponding author
    1. Department of Medical Physics and Clinical Engineering, Cambridge University Hospitals NHS Foundation Trust, Box 152, Addenbrookeˈs Hospital, Hills Road, Cambridge CB2 0QQ, United Kingdom and Department of Oncology, Cambridge University, Cambridge CB2 0QQ, United Kingdom
    Search for more papers by this author
  • Harden Susan V.,

    1. Oncology Centre, Cambridge University Hospitals NHS Foundation Trust, Box 192, Addenbrookeˈs Hospital, Hills Road, Cambridge CB2 0QQ, United Kingdom
    Search for more papers by this author
  • Thomas Simon J.

    1. Department of Medical Physics and Clinical Engineering, Cambridge University Hospitals NHS Foundation Trust, Box 152, Addenbrookeˈs Hospital, Hills Road, Cambridge CB2 0QQ, United Kingdom and Department of Oncology, Cambridge University, Cambridge CB2 0QQ, United Kingdom
    Search for more papers by this author

Abstract

Purpose:

Dose differences from those planned can occur due to the respiratory interplay effect on helical tomotherapy. The authors present a technique to calculate single-fraction doses in three-dimensions resulting from craniocaudal motion applied to a patient CT set. The technique is applied to phantom and patient plans using patient respiratory traces. An additional purpose of the work is to determine the contribution toward the interplay effect of different components of the respiratory trace.

Methods:

MATLAB code used to calculate doses to a CT dataset from a helical tomotherapy plan has been modified to permit craniocaudal motion and improved temporal resolution. Real patient traces from seven patients were applied to ten phantom plans of differing field width, modulation factor, pitch and fraction dose, and simulations made with peak-to-peak amplitudes ranging from 0 to 2.5 cm. PTV voxels near the superior or inferior limits of the PTV are excluded from the analysis. The maximum dose discrepancy compared with the static case recorded along with the proportion of voxels receiving more than 10% and 20% different from prescription dose. The analysis was repeated with the baseline variation of the respiratory trace removed, leaving the cyclic component of motion only. Radiochromic film was used on one plan-trace combination and compared with the software simulation. For one case, filtered traces were generated and used in simulations which consisted only of frequencies near to particular characteristic frequencies of the treatment delivery. Intraslice standard deviation of dose differences was used to identify potential MLC interplay, which was confirmed using nonmodulated simulations. Software calculations were also conducted for four realistic patient plans and modeling movement of a patient CT set with amplitudes informed by the observed motion of the GTV on 4DCT.

Results:

The maximum magnitude of dose difference to a PTV voxel due to the interplay effect within a particular plan-trace combination for peak-to-peak amplitudes of up to 2.5 cm ranged from 4.5% to 51.6% (mean: 23.8%) of the dose delivered in the absence of respiratory motion. For cyclic motion only, the maximum dose differences in each combination ranged from 2.1% to 26.2% (mean: 9.2%). There is reasonable correspondence between an example of the phantom plan simulations and radiochromic film measurement. The filtered trace simulations revealed that frequencies close to the characteristic frequency of the jaw motion across the target were found to generate greater interplay effect than frequencies close to the gantry frequency or MLC motion. There was evidence of interplay between respiratory motion and MLC modulation, but this is small compared with the interplay between respiratory motion and jaw motion. For patient-plan simulations, dose discrepancies are seen of up to 9.0% for a patient with 0.3 cm peak-to-peak respiratory amplitude and up to 17.7% for a patient with 0.9 cm peak-to-peak amplitude. These values reduced to 1.3% and 6.5%, respectively, when only cyclic motion was considered.

Conclusions:

Software has been developed to simulate craniocaudal respiratory motion in phantom and patient plans using real patient respiratory traces. Decomposition of the traces into baseline andcyclic components reveals that the large majority of the interplay effect seen with the full trace is due to baseline variation during treatment.

Ancillary