计算机断层显像与单光子发射型计算机断层肺灌注显像配准方法的研究

A method for registration between CT and single photon emission computed tomography lung perfusion images

目的比较目测法与不同个数和组合的外标记点法图像配准的差异，总结出一套精确度高、重复性好的计算机断层显像（CT）与单光子发射型计算机断层（SPECT）肺灌注图像配准的方法。方法选取两例肺癌患者分别行胸部CT和SPECT肺灌注扫描，扫描前分别于胸部体表黏贴9个标记点作为图像配准的参考点。分别依据目测法、2—9点法图像配准，记录配准后两图像9个外标记点三维空间坐标，以配准后外标记点在SPECT肺灌注图像的空间坐标值与在CT图像的空间坐标值差的绝对值作为图像配准精度的观察指标。结果采用9点法图像配准误差在三维方向上，以左右方向最大，为（1．92±1．56）mm。与9点法相比，目测法、2点法、3点法及4点法配准误差有统计学意义（P〈0．05）；5点法、6点法、7点法及8点法配准误差异无统计学意义（P〉0．05）。比较以中胸部标记点为基础的选点法与以下胸部标记点为基础的选点法配准差异，两者之间、前者与9点法之间差异均无统计学意义（P〉0．05），而后者与9点法之间差异有统计学意义（P=0．01）。5点法三次配准间差异无统计学意义（P〉0．05）。结论5点法图像配准是精确度高、重复性好的CT—SPECT异机图像配准方法；以受呼吸运动影响较小的中胸部外标记点为基础选取标记点，是较好的选点方法。

Objective To compare the registration errors of single photon emission computed tomography （SPECT） and CT images using a visual method and the methods of different numbers and combinations of the external markers, and to explore an accurate, reproducible and simple registration method of CT-SPECT lung perfusion images. Methods Two patients with lung cancer underwent thoracic CT scans and SPECT lung perfusion scans. The 9 external markers were fixed on the thoracic region before CT scans and SPECT lung perfusion scans. Two series of images were registered using a visual method and the methods of different numbers and combinations of the external markers respectively based on Pinnacle^3 planning treatment system. After registration, space coordinates of the 9 external markers were recorded respectively. The accuracy of registration was assessed using the absolute differences value of the 9 external markers coordinates between two series of images after registration. Results After registration using 9 external markers, the error in the left-right direction was the greatest, （ 1.92 ± 1.56） mm. Compared with that of using 9 external markers, there were statistical differences using the visual method, 2 external markers, 3 external markers, 4 external markers （ P 〈 0. 05 ）, and there was no statistical difference using 5 external makers, 6 external markers, 7 external markers, 8 external markers （P =0. 081,0. 061,0. 854,0. 860, respectively）. Comparing the registration methods based on the middle thoracic external markers with the registration methods based on the pectus external markers using 4 external markers,5 external markers,6 external markers, there was no statistical difference between two methods and between the former and 9 external markers （ P = 0. 395,0. 110, respectively）, and there was statistical difference between the latter and 9 external markers （ P = 0. 010）. There was no statistical difference in the three registrations using 5 external markers （ P = 0. 613,0. 26,0. 531, respectively）. Conclusion The method of 5 external markers distributing non-coplanar is an accurate, reproducible and simple registration method of CT-SPECT lung perfusion images. Choosing the middle thoracic external markers less influenced by respiratory movement is a better method.