King M A, Xia W, deVries D J, Pan T S, Villegas B J, Dahlberg S, Tsui B M, Ljungberg M H, Morgan H T
Department of Nuclear Medicine, University of Massachusetts Medical Center, Worcester 01655, USA.
J Nucl Cardiol. 1996 Jan-Feb;3(1):18-29. doi: 10.1016/s1071-3581(96)90020-3.
Significant hepatobiliary accumulation of technetium 99m-labeled cardiac perfusion agents has been shown to cause alterations in the apparent localization of the agents in the cardiac walls. A Monte Carlo study was conducted to investigate the hypothesis that the cardiac count changes are due to the inconsistencies in the projection data input to reconstruction, and that correction of the causes of these inconsistencies before reconstruction, or including knowledge of the physics underlying them in the reconstruction algorithm, would virtually eliminate these artifacts.
The SIMIND Monte Carlo package was used to simulate 64 x 64 pixel projection images at 128 angles of the three-dimensional mathematical cardiac-torso (MCAT) phantom. Simulations were made of (1) a point source in the liver, (2) cardiac activity only, and (3) hepatic activity only. The planar projections and reconstructed point spread functions (PSFs) of the point source in the liver were investigated to study the nature of the inconsistencies introduced into the projections by imaging, and how these affect the distribution of counts in the reconstructed slices. Bull's eye polar maps of the counts at the center of the left ventricular wall of filtered back-projection (FBP) and maximum-likelihood expectation-maximization (MLEM) reconstructions of projections with solely cardiac activity, and with cardiac activity plus hepatic activity scaled to have twice the cardiac concentration, were compared to determine the magnitude and location of apparent changes in cardiac activity when hepatic activity is present. Separate simulations were made to allow the investigation of stationary spatial resolution, distance-dependent spatial resolution, attenuation, and scatter. The point source projections showed significant inconsistencies as a function of projection angle with the largest effect being caused by attenuation. When consistent projections were simulated, no significant impact of hepatic activity on cardiac counts was noted with FBP, or 100 iterations of MLEM. With inconsistent projections, reconstruction of 180 degrees resulted in greater apparent cardiac count losses than did 360 degrees reconstruction for both FBP and MLEM. The incorporation of attenuation correction in MLEM reconstruction reduced the changes in cardiac counts to that seen in simulations in which attenuation was not included, but resulted in increased apparent localization of activity in the posterior wall of the left ventricle when scatter was present in the simulated images.
The apparent alterations in cardiac counts when significant hepatic localization is present is due to the inconsistency of the projections inherent in imaging. Prior correction of these, or accounting for them in the reconstruction algorithm, will virtually eliminate them as causes of artifactual changes in localization. Attenuation correction and scatter correction are both required to overcome the major sources of apparent count changes in the heart associated with hepatic uptake.
已表明锝99m标记的心脏灌注剂在肝胆中大量蓄积会导致这些药剂在心脏壁中的表观定位发生改变。进行了一项蒙特卡洛研究，以探究以下假设：心脏计数变化是由于输入重建的投影数据不一致所致，并且在重建前纠正这些不一致的原因，或者在重建算法中纳入其背后的物理知识，实际上将消除这些伪影。
使用SIMIND蒙特卡洛软件包在128个角度对三维数学心脏-躯干（MCAT）体模进行64×64像素投影图像模拟。模拟了以下三种情况：（1）肝脏中的点源；（2）仅心脏活动；（3）仅肝脏活动。研究了肝脏中点源的平面投影和重建的点扩散函数（PSF），以研究成像引入投影中的不一致的性质，以及这些如何影响重建切片中的计数分布。比较了仅具有心脏活动以及心脏活动加肝脏活动（肝脏活动按比例缩放至心脏浓度的两倍）的投影经滤波反投影（FBP）和最大似然期望最大化（MLEM）重建后左心室壁中心处计数的靶心极坐标图，以确定存在肝脏活动时心脏活动表观变化的幅度和位置。进行了单独的模拟，以研究固定空间分辨率、距离依赖性空间分辨率、衰减和散射。点源投影显示出作为投影角度函数的显著不一致，其中衰减造成的影响最大。当模拟一致的投影时，FBP或MLEM的100次迭代未发现肝脏活动对心脏计数有显著影响。对于不一致的投影，180度重建导致的心脏计数表观损失比FBP和MLEM的360度重建更大。在MLEM重建中纳入衰减校正可将心脏计数的变化减少到不包括衰减的模拟中所见的水平，但当模拟图像中存在散射时，会导致左心室后壁活动的表观定位增加。
当存在显著的肝脏定位时，心脏计数的表观改变是由于成像中固有的投影不一致所致。事先对这些进行校正，或在重建算法中考虑这些因素，实际上将消除它们作为定位中伪影变化的原因。需要同时进行衰减校正和散射校正，以克服与肝脏摄取相关的心脏中表观计数变化的主要来源。
