Slomka Piotr J, Nishina Hidetaka, Berman Daniel S, Kang Xingping, Akincioglu Cigdem, Friedman John D, Hayes Sean W, Aladl Usaf E, Germano Guido
Department of Imaging, Cedars-Sinai Medical Center, Los Angeles, California 90048, USA.
J Nucl Med. 2004 Jul;45(7):1128-34.
Gated myocadial perfusion SPECT (MPS) incorporates functional and perfusion information of the left ventricle (LV). To improve the image quality and accuracy of gated MPS we propose to eliminate the influence of cardiac LV motion in the display and quantification by a novel "motion-frozen" (MF) technique.
Three-dimensional LV contours were identified on images of the individual time phases. Three-dimensional phase-to-phase motion vectors were derived by sampling of the epi- and endocardial surfaces. A nonlinear image warping (thin-plate spline) was applied to warp all image phases to fit the end-diastolic (ED) phase. Warped images were created to provide the LV image in the ED phase but containing counts from an arbitrary number of time intervals. MF quantification has been performed using the same phase-to-phase motion vectors. MF normal perfusion limits were created from (99m)Tc sestamibi gated MPS studies of 40 females and 40 males with low likelihood (<5%) of coronary artery disease. All MF processing was completely automated. In the initial evaluation, we assessed the display quality and quantification of stress images using MF processing in 51 consecutive patients with 16-frame electrocardiographic gating and available coronary angiography.
The display quality was significantly better for MF images as assessed visually. The MF images had the appearance of ED frames but were less noisy and of higher resolution than the summed images. MF images had higher maximum count values in the LV (116% +/- 6%) and higher contrast (12.5 +/- 7.7 vs. 9.5 +/- 3.2) than the corresponding summed images. The area under the receiver operator characteristic curve for prediction of stenoses > or = 70% by the MF method was 0.92 +/- 0.04 versus 0.89 +/- 0.04 by standard quantification (P = not significant). The computation time for automated MF quantification and warping was <25 s for each case.
We have developed a novel technique for display and quantification of gated myocardial perfusion images, which retrospectively eliminates blur due to cardiac motion. Such processing of gated MPS appears to improve the effective resolution of images. Initial evaluation indicates that it may improve the accuracy of gated MPS in detection of coronary artery disease.
门控心肌灌注单光子发射计算机断层扫描(MPS)结合了左心室(LV)的功能和灌注信息。为了提高门控MPS的图像质量和准确性,我们建议通过一种新颖的“运动冻结”(MF)技术来消除心脏左心室运动在显示和定量分析中的影响。
在各个时间相的图像上识别三维左心室轮廓。通过对心外膜和心内膜表面进行采样得出三维逐相运动向量。应用非线性图像变形(薄板样条)将所有图像相进行变形,以使其与舒张末期(ED)相匹配。创建变形后的图像以提供处于ED相的左心室图像,但包含来自任意数量时间间隔的计数。使用相同的逐相运动向量进行MF定量分析。MF正常灌注限值是根据40名女性和40名冠心病可能性较低(<5%)的男性的(99m)Tc司他米比门控MPS研究得出。所有MF处理均完全自动化。在初始评估中,我们对51例连续患者进行了16帧心电图门控且有可用冠状动脉造影的研究,使用MF处理评估了负荷图像的显示质量和定量分析。
经视觉评估,MF图像的显示质量明显更好。MF图像具有ED帧的外观,但比叠加图像噪声更小且分辨率更高。MF图像在左心室中的最大计数值更高(116%±6%),对比度也更高(12.5±7.7对9.5±3.2),高于相应的叠加图像。MF方法预测狭窄≥70%的受试者操作特征曲线下面积为0.92±0.04,而标准定量分析为0.89±0.04(P = 无显著差异)。每个病例的自动MF定量分析和变形的计算时间<25秒。
我们开发了一种用于门控心肌灌注图像显示和定量分析的新技术,该技术可回顾性地消除由于心脏运动导致的模糊。这种门控MPS处理似乎提高了图像的有效分辨率。初始评估表明,它可能提高门控MPS检测冠状动脉疾病的准确性。