A novel dose-based positioning method for CT image-guided proton therapy

Abstract

Purpose: Proton dose distributions can potentially be altered by anatomical changes in the beam path despite perfect target alignment using traditional image guidance methods. In this simulation study, the authors explored the use of dosimetric factors instead of only anatomy to set up patients for proton therapy using in-room volumetric computed tomographic (CT) images.

Methods: To simulate patient anatomy in a free-breathing treatment condition, weekly time-averaged four-dimensional CT data near the end of treatment for 15 lung cancer patients were used in this study for a dose-based isocenter shift method to correct dosimetric deviations without replanning. The isocenter shift was obtained using the traditional anatomy-based image guidance method as the starting position. Subsequent isocenter shifts were established based on dosimetric criteria using a fast dose approximation method. For each isocenter shift, doses were calculated every 2 mm up to ± 8 mm in each direction. The optimal dose alignment was obtained by imposing a target coverage constraint that at least 99% of the target would receive at least 95% of the prescribed dose and by minimizing the mean dose to the ipsilateral lung.

Results: The authors found that 7 of 15 plans did not meet the target coverage constraint when using only the anatomy-based alignment. After the authors applied dose-based alignment, all met the target coverage constraint. For all but one case in which the target dose was met using both anatomy-based and dose-based alignment, the latter method was able to improve normal tissue sparing.

Conclusions: The authors demonstrated that a dose-based adjustment to the isocenter can improve target coverage and/or reduce dose to nearby normal tissue.

Figure 1

Planned dose distribution and beam angles for patient 1.

Figure 2

Workflow diagram for the dose-based alignment method compared to the current IGRT workflow.

Figure 3

Scatter plot and histogram of vector magnitude differences of the isocenter shifts comparing the dose-based alignments with the bony alignment and iGTV alignment.

Figure 4

DVHs for the target (left) and ipsilateral lung (right) for patient 13. The anatomy-based and dose-based alignment DVHs are shown by the solid and dashed lines, respectively. The DVH bands show the extent of all of the dose calculations in the search space, with the lighter gray band corresponding to the calculated DVHs that did not meet the target coverage constraint and the darker gray bands showing the DVHs that did meet the constraint.

Figure 5

An objective plot of the solution space of mean ipsilateral lung dose versus target D_99%. Calculated solutions are shown in the dots, with the doses for anatomy-based and optimized single-isocenter dose-based alignments shown in the red (lower) and green (upper) stars, respectively. The target coverage constraint is shown by the green dashed line (above the line is meeting the constraint), and the Pareto frontier (nondominated solutions) is shown by the solid red lines.

Figure 6

Dose distributions for patient 13 for the (a) plan and the last weekly CT based on (b) bony-alignment, and (c) dose-alignment. The dose clouds between 95% and 110% prescription dose levels for the bony-alignment and dose-alignment methods are also presented in axial, coronal, sagittal viewing planes to illustrate target coverage and locations of underdosage.