Geometric misalignments in fixed-portal intensity modulated radiation therapy

Purpose: Intensity modulated radiation therapy (IMRT) is being investigated to provide dose distributions that deliver an adequate dose to the target volume while sparing neighboring critical structures. The fluence distribution of an optimized plan may be very inhomogeneous, with local regions of high dose gradient (>10% mm-1). Even a small misalignment of one portal may cause large errors in the dose distribution. We have determined the effects of collimator, couch, and gantry angle misalignments on IMRT-generated dose distributions. Effects of translational errors were not addressed. Materials and Methods: The dose distribution D within the patient was assumed to be of the form: (D=P∫Ψk) where Ψ is the photon fluence, k is a two-dimensional dose-spread kernel, and P represents the depth-behavior of the dose distribution. The error in the dose distribution ΔD due to an offset Δx→ in the dose calculation position is, to first order, The error in the dose distribution was evaluated for each of the collimator, gantry, and couch rotation errors. The fluence gradient was determined in the collimator coordinate system, and Δx→ was translated into that system. Results: The error in the dose distribution is a function of the dose gradient and the distance of the dose calculation point to the axis of rotation. For example, with a collimator angle misalignment of ΔΘc, the error in dose for a point (xc,yc,zc) described in the collimator coordinate system (z-axis coincident with the beam central axis) was equal to where ρ=√x2c + y2c, αc=tan-1(yc(xc)). The depth-dose (P contribution to the gradient was assumed to be small relative to the fluence gradient. The dose error due to unintended gantry and couch misalignment was also determined. The error in dose was large for moderate dose gradients and off-axis distances for even a small angle setting error. For example, an error in the setup of the collimator angle of ΔΘc = 1 deg. at a point with off-axis distance ρ = 7cm and lateral dose gradient vertical bar ∇-vector D vertical bar =10% mn-1 results in a dose error of more than 10%. Conclusions: The application of IMRT in clinical dose delivery will require the improved quality assurance of the accelerator angle setting readouts, as well as the application of fluence gradient limitations to assure that the planned dose is clinically indistinguishable from the delivered dose