Target-tracking deliveries using conventional multileaf collimators planned with 4D direct-aperture optimization.

Respiratory motion-induced degradation of intensity-modulated radiotherapy can be corrected by the dynamic target-tracking motion of multileaf collimator equipment on a conventional linear accelerator. This paper presents a new system by which the motion of the tissue and the delivery equipment can be incorporated into the treatment optimization using a 4D direct-aperture optimization method. The program can optimize a static or dynamic delivery with respect to a 4D patient model. The individualized patient model consists of a series of discrete phases and describes changes in tissue: deformation, geometry, attenuation and scatter properties over the breathing cycle. A set of treatment apertures is matched to the respiratory phases of the motion model, and motion of the apertures between phases is constrained by the maximum leaf velocity. Plans with dynamic and static deliveries optimized on 4D patient models were compared to static plans optimized on a single phase. This investigation was carried out on a 4D digital motion phantom and repeated on a 4D patient model. The effect of motion of the static plan on the 4D phantom was evaluated by recalculating dose from all phases of the 4D model. The plan cost was evaluated as a combination of the rms spread in tumour dose from the prescribed dose and the volume of normal lung receiving doses above 10 Gy with relative weightings of 5 and 1 respectively. The motion was found to degrade the static plan by 30 +/- 4% with respect to the 3D cost function value. In contrast, the motion did not cause significant degradation to a treatment if the treatment was optimized on the 4D phantom and the cost was improved by 16 +/- 3% by optimizing with dynamic leaf tracking motion. All results are relative to the static single-phase plan. In the 4D patient model the observed tissue motion was considerably less and the measured benefit of 4D planning was consequently reduced. For the 4D patient the plan cost was not significantly changed by the tissue motion. Optimizing on 4D patient conferred an improvement of 7.5 +/- 0.3%, and the 4D plan with dynamic leaf motion improved the plan cost by 8.3 +/- 0.1%.