Initial Clinical Experience with Using an Automated Dose Tracking Tool to Verify Dose Delivery and Trigger Adaptive Replanning

Abstract

We report on our initial clinical use of an automated software tool which incorporates deformable registration and dose summation to identify and track inter-fraction anatomic changes and delivered doses based upon daily MVCT. We used this tool to verify dose delivery and to trigger adaptive replanning. For the first 32 patients (12 male pelvis, 9 female pelvis, 8 head and neck, 3 pediatric) treated on our helical tomotherapy system, daily MVCT IGRT images and MVCT-to-plan-CT shift information were input to the dose-tracking software tool. For each fraction, the tool calculates a deformable registration map to warp each region-of-interest (ROI) from the planning CT to the daily MVCT, and recalculates the “delivered” dose on the MVCT based on the planned beam configuration (sinogram). Changes to the ROI volumes and dose for each fraction are tracked, and the dose distribution is summed for all fractions to date. The planned DVH is compared with the delivered DVH in terms of accumulated dose for all fractions delivered to date, as well as projected dose (which assumes that the dose distribution for the most-recent fraction applies to the remaining fractions). The tool also enables review of image registration, ROI deformation, and dose recalculation for each fraction. Aside from initial enrollment of the patient’s plan within the software tool, no further user intervention is required. Tests of the software tool upon a cylindrical solid-water phantom demonstrated that the MVCT-based dose distributions are consistent with the planned dose to within 1%. For several delivered pelvic radiotherapy plans, review of the ROIs and delivered dose distribution on the daily MVCT verified that the CTV coverage was achieved for each fraction, although for some fractions underdosing of the peripheral PTV was observed and confirmed by the accumulated DVH comparison with the plan. PTV underdosing was attributed to PTV warping uncertainties due to its location within soft tissue not bordering obvious anatomic landmarks, or (as in the case of one prostate patient) to anatomic changes within the patient such as a 1 to 2 cm increase of the anterior extent. For one head and neck patient, for whom weight loss was suspected due to degraded fit of the immobilization mask, the software tool revealed progressive geometric miss of the PTV over a one-week time frame; based on the delivered PTV DVHs, which indicated the dose covering 95% of the PTV had decreased by 5 cGy from the original planned dose of 175 cGy per fraction, an adaptive replan was advised for this patient. Our initial clinical use of the automated dose-tracking tool for tomotherapy delivery indicates that the tool can effectively verify the delivery of planned target coverage and organ-at-risk sparing, and can identify and confirm sufficient degradation of plan quality to trigger adaptive replanning.