Wang He, Dong Lei, Lii Ming Fwu, Lee Andrew L, de Crevoisier Renaud, Mohan Radhe, Cox James D, Kuban Deborah A, Cheung Rex
Department of Radiation Physics, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030, USA.
Int J Radiat Oncol Biol Phys. 2005 Mar 1;61(3):725-35. doi: 10.1016/j.ijrobp.2004.07.677.
Daily prostate deformation hinders accurate calculation of dose, especially to intraprostatic targets. We implemented a three-dimensional deformable registration algorithm to aid dose tracking for targeted prostate radiotherapy.
The algorithm registers two computed tomography (CT) scans by iteratively minimizing their differences in image intensity. For validation, we measured the accuracy in registering (a) a pelvic CT set to its mathematically deformed counterpart, (b) CT scans of a deformable pelvic phantom with and without an endorectal balloon inflated, to simulate intraprostatic targets, 23 CT-opaque seeds were embedded in the prostate, and (c) two pelvic CT scans of a patient obtained on 2 separate days.
The mean (SD) error in registering the pelvic CT set to its transformed set was 0.5 mm (1.5), with correlation coefficient improvement from 0.626 to 0.991. Using the deformable pelvic phantom, the correlation coefficient improved from 0.543 to 0.816 after registration. The mean (SD) error in tracking the intraprostatic seeds was 0.8 mm (0.5). The correlation coefficient improved from 0.610 to 0.944 after registration of the two patient CT sets.
The algorithm had an accuracy of about 1 mm. It could be used for optimizing dose calculation and delivery for prostate radiotherapy.
每日前列腺变形会妨碍剂量的精确计算,尤其是对于前列腺内靶区的剂量计算。我们实施了一种三维可变形配准算法,以辅助靶向前列腺放疗的剂量追踪。
该算法通过迭代最小化两幅计算机断层扫描(CT)图像的强度差异来对其进行配准。为进行验证,我们测量了以下配准的准确性:(a)将盆腔CT数据集与其数学变形后的对应数据集进行配准;(b)对一个可变形盆腔模体在直肠内气囊充气和未充气的情况下进行CT扫描,以模拟前列腺内靶区,在前列腺中植入23颗CT显影的种子;(c)对一名患者在两个不同日期获得的两幅盆腔CT扫描图像进行配准。
将盆腔CT数据集与其变换后的数据集进行配准时,平均(标准差)误差为0.5毫米(1.5),相关系数从0.626提高到0.991。使用可变形盆腔模体时,配准后相关系数从0.543提高到0.816。追踪前列腺内种子的平均(标准差)误差为0.8毫米(0.5)。对两名患者的CT数据集进行配准后,相关系数从0.610提高到0.944。
该算法的准确性约为1毫米。它可用于优化前列腺放疗的剂量计算和投送。