Department of Medical Physics, Addenbrooke's Hospital, Cambridge, England.
Int J Radiat Oncol Biol Phys. 2012 Apr 1;82(5):e717-23. doi: 10.1016/j.ijrobp.2011.11.009. Epub 2012 Feb 11.
When one is comparing two dose distributions, a number of methods have been published to combine dose difference and distance to agreement into a single measure. Some have been defined as pass/fail indices and some as numeric indices. We show that the pass/fail indices can all be used to derive numeric indices, and we compare the results of using these indices to evaluate one-dimensional (1D) and three-dimensional (3D) dose distributions, with the aim of selecting the most appropriate index for use in different circumstances.
The indices compared are the gamma index, the kappa index, the index in International Commission on Radiation Units & Measurements Report 83, and a box index. Comparisons are made for 1D and 3D distributions. The 1D distribution is chosen to have a variety of dose gradients. The 3D distribution is taken from a clinical treatment plan. The effect of offsetting distributions by known distances and doses is studied.
The International Commission on Radiation Units & Measurements Report 83 index causes large discontinuities unless the dose gradient cutoff is set to equal the ratio of the dose tolerance to the distance tolerance. If it is so set, it returns identical results to the kappa index. Where the gradient is very high or very low, all the indices studied in this article give similar results for the same tolerance values. For moderate gradients, they differ, with the box index being the least strict, followed by the gamma index, and with the kappa index being the most strict.
If the clinical tolerances are much greater than the uncertainties of the measuring system, the kappa index should be used, with tolerance values determined by the clinical tolerances. In cases where the uncertainties of the measuring system dominate, the box index will be best able to determine errors in the delivery system.
在比较两种剂量分布时，已经发表了许多方法将剂量差异和到一致的距离结合到一个单一的度量中。其中一些已被定义为通过/失败指数，而另一些则被定义为数字指数。我们表明，通过/失败指数都可以用于导出数字指数，并且我们比较了使用这些指数来评估一维（1D）和三维（3D）剂量分布的结果，目的是选择最适合在不同情况下使用的指数。
比较的指数是伽玛指数、kappa 指数、国际辐射单位和测量委员会报告 83 中的指数和框指数。比较了 1D 和 3D 分布。选择 1D 分布以具有各种剂量梯度。3D 分布取自临床治疗计划。研究了通过已知距离和剂量偏移分布的效果。
国际辐射单位和测量委员会报告 83 指数会导致大的不连续性，除非将剂量梯度截止值设置为等于剂量公差与距离公差的比值。如果这样设置，它将返回与 kappa 指数相同的结果。在梯度非常高或非常低的情况下，本文研究的所有指数对于相同的公差值都给出相似的结果。对于中等梯度，它们会有所不同，框指数最不严格，其次是伽玛指数，而 kappa 指数最严格。
如果临床公差远大于测量系统的不确定性，则应使用 kappa 指数，公差值由临床公差确定。在测量系统的不确定性占主导地位的情况下，框指数将能够最好地确定输送系统中的误差。
