优化的CT图像配准方法的研究与比较

Investigation and Comparision of Optimal Registration for Medical CT Images

由于不同的医学图像信息的局限性,综合利用多次成像或多种模式成像可以获得较全面的信息,使临床诊断和治疗、放疗的定位和计划设计、外科手术和疗效评估等更加全面和精确。尽管近30年来对生物医学图像配准方法做了大量的研究,并提出了许多解决办法,取得了很大的进展,但到目前为止还没有一种能应用于临床的通用的软件配准方法。为了选出更合适的CT医学图像配准方法,试验与比较了一些配准方法。为了能处理更广范围的形变,特别地在CT医学图像上对多分辨率的弹性配准进行了研究。实验结果表明,对于基于B样条的弹性配准,多分辨率配准模式性能更优。

Due to its capability of maximizing the embedded information from a large amount of medical images captured from different imaging modalities and acquired from different scanning intervals,medical image registration plays an essential role in clinical applications including diagnosis,radiotherapy planning,assessment of patient＇s response to the treatment.An intensive research on registration methodologies has been conducted during the last three decades.There is no general software solution to clinically acceptable registration.To sort out more appropriate registration for clinical CT images,registration methods were tested and compared in this paper.The multiresolution deformable registration was investigated for accommodating a wider range of deformations.The experimental results on the clinical CT datasets demonstrated the superior performance of the B-spline deformable registration by multiresolution scheme.