Dosimetric Comparison of MLC Leaf Sequencing Techniques for Static and Dynamic Intensity Modulated Radiotherapy for Nasopharyngeal Carcinoma

Materials/Methods: Using current imaging technologies (e.g. cone beam CT, portal imaging), an image of the tumor target is easily obtained before and/or during each fraction of radiation. This allows the shape of the target, as seen in the beam’s-eye view, to be defined on sequential patient images. A rigid-body transformation is then applied to remove possible translation and rotation differences between the target shapes. Corresponding landmarks on each processed target shape are automatically identified using the MDL principle. Next, ASM, a statistical description of the target shape, is constructed based on the landmarks. In ASM the shape variability is learned through observation. Finally, a Monte Carlo method is used to generate synthetic target shapes based on ASM, as follows. Points will be sampled in the beam’s-eye view plane, and each point will be assigned a score based on the computed statistical probability that it is inside the target. The iso-score lines form level sets, with each level set corresponding to target shape and size for a given probability. These level sets reflect clinical information learned from the available images will be good candidates for radiotherapy field shapes with non-uniform margins.