Evaluation of multileaf collimator design for a photon beam.

Various aspects of multileaf collimator (MLC) design are examined relative to clinical requirements. The characteristics studied included: (a) irregular field edge definition or "effective" penumbra, (b) optimum field coverage for the multileaf portion of the field, and (c) leaf velocity. A film dosimetry technique was developed to measure the rapid 2-dimensional change in dose at an edge defined by a multileaf collimator with the segments staggered. The method applies a correction factor which allows for the changing ratio of scattered to primary photons at the field edge so that the energy dependence of the film is corrected. Stepped lead alloy blocks were irradiated with 6 MV photons to obtain films simulating a double-focused multileaf collimator, and the results were compared to films of fields shaped with standard divergent blocks. The effect of the shape of the leaf face (the end of the leaf) on penumbra was also studied. Proper shaping of the leaf ends may eliminate the need to exactly match beam divergence so that the mechanical of the collimator system is simplified. Leaves having several different end shapes and moving horizontally to intercept a vertical beam were compared to the divergent design where a straight face moves along an arc. The measurements showed that the "effective" penumbra (measured as the distance from the 80 to 20% isodose lines) for the multileaf collimator is a function of the angle between the direction of leaf motion and the edge defined by the leaves. In addition, all leaf end shapes showed some increase in penumbra compared to standard divergent blocking and also had increasing penumbra width as they moved over or back from the field center line. A total of 459 treatment fields and six disease sites were examined to determine the percentage of fields potentially shaped by multileaf segments of specified length. This study showed 93% of the fields had lengths of 30 cm or less and 99% had widths of 25 cm or less. A study conducted to determine the required leaf velocity to shape various target volume configurations during complete rotation (at 1 RPM) showed that a leaf speed of at least 1.5 cm/sec at isocenter is needed for dynamic conformal treatment.