Plattard D, Soret M, Troccaz J, Vassal P, Giraud J Y, Champleboux G, Artignan X, Bolla M
TIMC/IMAG Laboratory, Faculté de Médecine, Domaine de la Merci, Grenoble, France.
Comput Aided Surg. 2000;5(4):246-62. doi: 10.1002/1097-0150(2000)5:4<246::AID-IGS4>3.0.CO;2-Z.
Conformal radiation therapy requires accurate patient set-up for each fraction delivery. Electronic portal imaging devices allow the acquisition of portal images just before and even during dose delivery. However, the quantitative interpretation of these images in determining and correcting the patient's position remains uncertain, and automated methods are therefore being developed. Such methods must be usable for the different radiation therapy techniques. They must be robust and as automated as possible for use in clinical routines. This work was undertaken to establish the feasibility of 2D/2D registration for portal/portal and portal/simulator images in radiotherapy.
This paper describes an automated method based on the combination of calibration algorithms and pixel-based registration algorithms. We present experiments with the different imaging techniques, some of which use a phantom with and without a gold standard. Preliminary results obtained using patient data are also presented and discussed.
The results obtained with a phantom demonstrated that this automated method for 2D/2D registration is fast, accurate, and robust, even in the case of blurred images for small treatment fields.
Mutual information is a feasible method for 2D/2D portal/portal and portal/simulator image registration in radiotherapy.
适形放射治疗要求在每次分次照射时准确摆位患者。电子射野影像装置能够在剂量照射前甚至照射过程中采集射野图像。然而,在确定和校正患者位置时对这些图像进行定量解读仍不明确,因此正在开发自动化方法。此类方法必须适用于不同的放射治疗技术。它们必须稳健且尽可能自动化以便用于临床常规。开展这项工作是为了确定放射治疗中射野/射野和射野/模拟定位图像二维/二维配准的可行性。
本文描述了一种基于校准算法和基于像素的配准算法相结合的自动化方法。我们展示了针对不同成像技术的实验,其中一些实验使用了带有和不带有金标准的模体。还展示并讨论了使用患者数据获得的初步结果。
使用模体获得的结果表明,这种二维/二维配准的自动化方法快速、准确且稳健,即使对于小照射野的模糊图像也是如此。
互信息是放射治疗中二维/二维射野/射野和射野/模拟定位图像配准的一种可行方法。