Putzer Daniel, Henninger Benjamin, Kovacs Peter, Uprimny Christian, Kendler Dorota, Jaschke Werner, Bale Reto J
Department of Radiology, Innsbruck Medical University, Innsbruck, Austria -
Q J Nucl Med Mol Imaging. 2016 Jun;60(2):172-81.
Even as PET/CT provides valuable diagnostic information in a great number of clinical indications, availability of hybrid PET/CT scanners is mainly limited to clinical centers. A software-based image fusion would facilitate combined image reading of CT and PET data sets if hardware image fusion is not available. To analyze the relevance of retrospective image fusion of separately acquired PET and CT data sets, we studied the accuracy, practicability and reproducibility of three different image registration techniques.
We evaluated whole-body 18F-FDG-PET and CT data sets of 71 oncologic patients. Images were fused retrospectively using Stealth Station System, Treon (Medtronic Inc., Louisville, CO, USA) equipped with Cranial4 Software. External markers fixed to a vacuum mattress were used as reference for exact repositioning. Registration was repeated using internal anatomic landmarks and Automerge software, assessing accuracy for all three methods, measuring distances of liver representation in CT and PET with reference to a common coordinate system.
On first measurement of image fusions with external markers, 53 were successful, 16 feasible and 2 not successful. Using anatomic landmarks, 42 were successful, 26 feasible and 3 not successful. Using Automerge Software only 13 were successful. The mean distance between center points in PET and CT was 7.69±4.96 mm on first, and 7.65±4.2 mm on second measurement. Results with external markers correlate very well and inaccuracies are significantly lower (P<0.001) than results using anatomical landmarks (10.38±6.13 mm and 10.83±6.23 mm). Analysis revealed a significantly faster alignment using external markers (P<0.001).
External fiducials in combination with immobilization devices and breathing protocols allow for highly accurate image fusion cost-effectively and significantly less time, posing an attractive alternative for PET/CT interpretation when a hybrid scanner is not available.
尽管正电子发射断层显像/计算机断层扫描（PET/CT）在大量临床指征中能提供有价值的诊断信息，但混合型PET/CT扫描仪主要限于临床中心使用。如果没有硬件图像融合，基于软件的图像融合将有助于对CT和PET数据集进行联合图像解读。为分析单独采集的PET和CT数据集进行回顾性图像融合的相关性，我们研究了三种不同图像配准技术的准确性、实用性和可重复性。
我们评估了71例肿瘤患者的全身18F-氟代脱氧葡萄糖（18F-FDG）-PET和CT数据集。使用配备Cranial4软件的美国美敦力公司（Medtronic Inc.，路易斯维尔，科罗拉多州）的Stealth Station System、Treon进行回顾性图像融合。固定在真空床垫上的外部标记物用作精确重新定位的参考。使用内部解剖标志和自动合并（Automerge）软件重复配准，评估所有三种方法的准确性，参照共同坐标系测量CT和PET中肝脏影像的距离。
首次使用外部标记物进行图像融合测量时，53例成功，16例可行，2例不成功。使用解剖标志时，42例成功，26例可行，3例不成功。仅使用自动合并软件时，13例成功。PET和CT中心点之间的平均距离首次测量时为7.69±4.96毫米，第二次测量时为7.65±4.2毫米。使用外部标记物的结果相关性非常好，且误差显著低于使用解剖标志的结果（分别为10.38±6.13毫米和10.83±6.23毫米）（P<0.001）。分析显示使用外部标记物时配准明显更快（P<0.001）。
外部基准点与固定装置和呼吸方案相结合，能以高性价比且显著更短的时间实现高度精确的图像融合，在没有混合型扫描仪时，为PET/CT解读提供了有吸引力的替代方案。
