Margins to compensate for respiratory-induced mismatches between lung tumor and fiducial marker positions using four-dimensional computed tomography

Abstract

Background and purpose

Tumors and fiducial markers do not always exhibit synchronous motion across different respiratory phases, in a phenomenon called the target localization error (TLE). We determined the margin to compensate for the TLE using four-dimensional computed tomography (4D-CT).

Materials and methods

We analyzed data from 21 lung tumor patients with fiducial markers; 11 for TLE determination and 10 for validation. Shifted CT images were generated by aligning the centroids of the fiducial markers in the reference phase of 4D-CT with those in each respiratory phase, and the union of gross tumor volumes (GTVs) was determined (${GTV}_{union}^{shift}$). Conversely, variations in GTV centroids across the respiratory phases were calculated, and the 95th percentile of the root mean square error was defined as the TLE. Using this TLE, a GTV with an added TLE (${GTV}_{TLE}^{ref}$) was generated in the reference phase. Subsequently, a treatment plan assuming dynamic tumor tracking (DTT) was created for the planning target volume, derived by adding an isotropic 5 mm margin to ${GTV}_{TLE}^{ref}$, and the dose coverage for ${GTV}_{union}^{shift}$ was evaluated.

Results

The TLEs (standard deviations of the root mean square error) were 2.0 (0.8) mm, 2.1(0.7) mm, and 3.2 (1.1) mm in the left − right, anterior − posterior, and superior − inferior directions, respectively. A dosimetric evaluation revealed that ${GTV}_{union}^{shift}$ did not receive 100 % of the prescribed dose in four of 10 cases owing to artifacts.

Conclusion

The TLE can be compensated by adding an anisotropic margin to the GTV in the reference phase, a critical consideration in DTT.