NSCLC—SBRT中4D—CBCT图像配准方式研究

The application research about registration methods of 4D-CBCT in non-small cell lung cancer with SBRT

目的探讨在4D．CBCT扫描后，人工图像配准与灰度T＋R图像配准在NSCLC（NSCLC）SBRT中的临床应用，得到在NSCLC．SBRT疗程中最适合的配准方式。方法10例接受SBRT的NSCLC患者，在分次4D—CBCT扫描后，对扫描图像与定位cT图像分别进行人工与灰度T＋R配准。在患者分次治疗过程中，对配准图像分别计算最大互信息、Haunsdorff距离、平均对称距离，并行配对t检验差异。结果第1～3次SBRT时，灰度T＋R配准较人工配准的最大互信息平均提高8．8％（P=0．003，0．000，0．008），Haunsdorff距离平均减少21．1％（P=0．000，0．000，0．000），平均对称距离平均较少13．4％（P=0．000，0．002，0．018）；第6～8次SBRT时，人工配准较灰度T＋R配准的最大互信息平均提高19．3％（P=0．014，0．000，0．000），Haunsdorff距离平均减少17．6％（P=0．000，0．000，0．000），平均对称距离平均较少19．6％（P=0．001，0．000，0．000）。结论在NSCLC患者治疗前期。4D．CBCT后应用灰度T＋R进行图像配准较为合理，治疗后期应用人工配准较为合理。

Objective To investigate the artificial and gray T＋R 4D-CBCT image registration for non-small cell lung cancer （NSCLC） with SBRT. Methods 2D SBRT 10 patients with NSCLC, their 4D- CBCT scan images and reference CT images were registered by the methods of artificial and gray T ＋ R registration. The maximum mutual information （ MI ） , Haunsdorff distance and average symmetry distance （ASD） were calculated and were analyzed by paired-t test. Results In 1-3 fraction SBRT,the maximum MI for Gray T＋R registration was averaged improved 8. 8% （P = 0. 003,0. 000,0. 008 ） , Haunsdorff distance and ASD were averaged reduced 21.1% （ P= 0. 000,0. 000, 0. 000） and 13.4% （ P= 0. 000,0. 002, 0. 018） respectively,and they all had the statistical significance compared with artificial registration in the early period of treatment;in 6-8 fraction SBRT, compared with gray T＋R registration the maximum MI for artificial registration was averaged improved 19. 3% （ P = 0. 014,0. 000,0. 000 ） , Haunsdorff distance and ASD were averaged reduced 17.6% （ P = 0. 000,0. 000,0. 000 ） and 19. 6% （ P = 0. 001,0. 000,0. 000 ） respectively, and they also had the statistical significance. Conclusions Gray T＋R image registration was reasonable in early period of treatment and artificial image registration was available in later period of treatment.