Dosimetric evaluation of the conformation of the multileaf collimator to irregularly shaped fields

The goal of this study was to evaluate the dosimetric characteristics of geometric MLC prescription strategies and compare them to those of conventional shielding block.

Circular fields, square fields, and 12 irregular fields for patients with cancer of the head and neck, lung, and pelvis were included in this study. All fields were shaped using the MLC and conventional blocks. A geometric criterion was defined as the amount of area discrepancy between the MLC and the prescription outline. The "least area discrepancy" (LAD) of the MLC conformation was searched by selecting the collimator angle, meanwhile keeping a preselected position along the width of the leaf into the prescribed field. Five LAD conventions were studied. These included the LAD-0, LAD-1/3, LAD-1/2, and LAD-2/3 that inserted the leaves at the 0, 1/3, 1/2, and 2/3 of the leaf end into the prescription field, respectively. In addition, the LAD optimization was applied to the transecting (TRN) approach for leaf conformation that prescribed an equal area of overblocking and underblocking under each leaf. Film dosimetry was performed in a 20 cm polystyrene phantom at 10 cm depth 100 cm from source to axis distance (SAD) for both 6 and 18 MV photons with each of the above MLC conformations and conventional blocks. The field penumbra width, defined as the mean of the separation between the 20% and 80% isodose lines along the normal of the prescription field edge, was calculated using both the MLC and conventional block film dosimetry and compared. In a similar way, the d20 is defined as the mean separation between the 20% isodose line and the prescription field edge, and the d80 is defined as the mean separation between the 80% isodose line and the prescription field edge.

The field penumbra width for all MLC conventions was approximately 2 mm larger than that of the conventional block. However, there was a larger variation of the separation distribution in the penumbra region of the irregular fields for the MLC, which had a standard deviation of 1 mm (a factor of 5 larger than the conventional block). The dosimetry for the circular fields showed that the LAD-TRN, LAD-1/2, and LAD-1/3 approximated the conventional blocking well in terms of d20 and d80; however, no single convention produced the best conformation for both measures. The dosimetric result of the patient treatment fields was similar for all sites. The LAD-1/3, LAD-1/2, and LAD-TRN strategies conformed to within 1 to 1.5 mm of the d80 of the conventional block for both 6 MV and 18 MV, respectively. The LAD-1/2 and LAD-TRN conformations were virtually identical, although it is proven analytically that the LAD-1/2 convention has the least overall area discrepancy of all conventions.

The five MLC conformation conventions resulted in similar dosimetric penumbrae for all field shapes studied. The LAD-1/3, LAD-TRN, and LAD-1/2 produced the more favorable approximation to conventional block. The field penumbra width, although useful for evaluating irregular field shapes, could not describe the large local variations in the penumbra along the field edge for the MLC. These local variations could be of clinical concern when they appear near vital organs. However, the variation in a local region can potentially be reduced by minimizing the jaggedness of the leaf steps in that local region. The dosimetric results were useful as guidelines for the clinicians in the evaluation and adjustment of MLC leaf positions.