基于Demons算法的弹性配准在放疗锥形束CT图像配准中的应用

Application of Elastic Registration Based on Demons Algorithm in Cone Beam CT

将基于经典Demons算法与加速Demons算法的弹性配准模型应用到放疗患者不同分次间的锥形束CT(CBCT)图像中,为进一步分析肿瘤及危及器官的变化提供软件支持。通过Matlab软件编写三维弹性配准程序,并用此程序对宫颈癌放疗患者不同分次时的两组锥形束CT图像进行仿真实验验证。结果表明经典Demons算法配准前后对比,最小均方误(MSE)减少59.7%,相关系数(CC)提高了11.0%;加速Demons算法配准前后对比,MSE减少40.1%,CC提高7.2%。实验验证上述两种基于Demons模型的弹性配准在CBCT图像配准中取得较好的配准效果,但对细微差别处仍显得精度不够,且整个配准时间较长,如要应用到在线自适应放疗中仍需进一步提高形变的精度并减少配准时间。

We applied Demons and accelerated Demons elastic registration algorithm in radiotherapy cone beam CT （CBCT） images, We provided software support for real-time understanding of organ changes during radiotherapy. We wrote a 3D CBCT image elastic registration program using Matlab software, and we tested and verified the images of two patients with cervical cancer 3D CBCT images for elastic registration, based on the classic Demons algorithm, minimum mean square error （MSE） decreased 59.7%, correlation coefficient （CC） increased 11.0%. While for the accelerated demons algorithm, MSE decreased 40. 1%, CC increased 7.2%. The experimental verification with two methods of demons algorithm obtained the desired results, but the small difference appeared to be lack of precision, and the total registration time was a little long. All these problems need to be further improved for accuracy and reducing of time.