Controlling Beam Complexity in Intensity Modulated Radiation Therapy.

Abstract

External beam intensity modulated radiation therapy (IMRT) is a technique in which the spatial intensity of radiation from each beam direction can be modulated to provide superior conformality of dose to a tumor volume while sparing important normal tissues. A fundamental and potentially limiting feature of IMRT is the highly complex fields that can be created through inverse plan optimization. Highly modulated treatments are a large departure from conventional radiotherapy methods, are difficult to deliver accurately and efficiently, and can result in an undesirable increase in leakage dose being delivered to the patient. Longer deliveries may also increase the chance for patient motion during treatment and could potentially reduce the probability of controlling some tumors. The large intensity fluctuations observed in IMRT beams are often a result of the degeneracy of the optimization problem, and the types of optimization method and cost function used. This work demonstrates that beam complexity is a result of these two issues, and is dependent on the placement of dose evaluation points in the target and normal tissues. This research shows that (i) optimizing surfaces instead of discrete beamlet intensities to represent the beam can reduce the degrees of freedom in IMRT and results in much smoother beams at the expense of a slight increase in normal tissues, (ii) maximum beamlet intensity restrictions are useful for improved delivery efficiency, but may restrict the optimizer at low limits, and (iii) modulation penalties can be incorporated into the cost function to promote plan smoothness without sacrificing plan quality. Penalizing the overall plan modulation is an effective way to reduce modulation, but it falsely penalizes the desirable beam modulation as well as the undesirable modulation. To address this problem, diffusion principles are used to develop a spatially adaptive smoothing method that only penalizes the unnecessary beam modulation and can be used without degrading plan quality. This method is customizable to a variety of treatment scenarios. The clinical impact of reducing beam complexity is significant, as it can result in an improvement in delivery accuracy and efficiency, quicker optimization times, and increased robustness to point sampling and geometric uncertainty.