Department of Physics and Astronomy, University of British Columbia, Vancouver, Canada.
Med Phys. 2011 Jul;38(7):4186-95. doi: 10.1118/1.3595112.
Normal patient databases (NPDs) are used to distinguish between normal and abnormal perfusion in SPECT myocardial perfusion imaging (MPI) and have gained wide acceptance in the clinical environment, yet there are limitations to this approach. This study introduces a template-based method for semi-quantitative MPI, which attempts to overcome some of the NPD limitations.
Our approach involves the construction of a 3D digital healthy heart template from the delineation of the patient's left ventricle in the SPECT image. This patient-specific template of the heart, filled with uniform activity, is then analytically projected and reconstructed using the same algorithm as the original image. Subsequent to generating bulls-eye maps for the patient image (PB) and the template image (TB), a ratio (PB/TB) is calculated, which produces a reconstruction-artifact corrected image (CB). Finally, a threshold is used to define defects within CB enabling measurements of the perfusion defect extent (EXT). The SPECT-based template (Ts) measurements were compared to those of a CT-based "ideal" template (TI). Twenty digital phantoms were simulated: male and female, each with one healthy heart and nine hearts with various defects. Four physical phantom studies were performed modeling a healthy heart and three hearts with different defects. The phantom represented a thorax with spine, lung, and left ventricle inserts. Images were acquired on General Electric's (GE) Infinia Hawkeye SPECT/CT camera using standard clinical MPI protocol. Finally, our method was applied to 14 patient MPI rest/stress studies acquired on the GE Infinia Hawkeye SPECT/CT camera and compared to the results obtained from Cedars-Sinai's QPS software.
In the simulation studies, the true EXT correlated well with the TI (slope= 1.08; offset = -0.40%; r = 0.99) and Ts (slope = 0.90; offset = 0.27%; r = 0.99) methods with no significant differences between them. Similarly, strong correlations were measured for EXT obtained from QPS and the template method for patient studies (slope =0.91; offset = 0.45%; r = 0.98). Mean errors in extent for the Ts method using simulation, physical phantom, and patient data were 2.7% +/- 2.4%, 0.9% +/- 0.5%, 2.0% +/- 2.7%, respectively.
The authors introduced a method for semi-quantitative SPECT MPI, which offers a patient-specific approach to define the perfusion defect regions within the heart, as opposed to the patient-averaged NPD methodology.
正常患者数据库（NPD）用于区分 SPECT 心肌灌注成像（MPI）中的正常和异常灌注，已在临床环境中得到广泛认可，但这种方法存在局限性。本研究引入了一种基于模板的半定量 MPI 方法，旨在克服 NPD 的一些局限性。
我们的方法涉及从 SPECT 图像中描绘患者左心室构建 3D 数字健康心脏模板。然后，使用与原始图像相同的算法对充满均匀活性的心脏进行分析，并对其进行投影和重建。在为患者图像（PB）和模板图像（TB）生成牛眼图之后，计算 PB/TB 比值，生成重建伪影校正图像（CB）。最后，使用阈值定义 CB 内的缺陷，从而可以测量灌注缺陷范围（EXT）。基于 SPECT 的模板（Ts）测量值与基于 CT 的“理想”模板（TI）的测量值进行了比较。模拟了 20 个数字体模：男性和女性，每个体模均有一个健康心脏和九个具有不同缺陷的心脏。进行了四项物理体模研究，模拟了一个具有不同缺陷的健康心脏和三个心脏。体模代表了带有脊柱、肺和左心室插件的胸部。使用标准的临床 MPI 协议，在通用电气（GE）的 Infinia Hawkeye SPECT/CT 相机上获取图像。最后，将我们的方法应用于在 GE Infinia Hawkeye SPECT/CT 相机上获取的 14 例患者 MPI 静息/负荷研究，并将结果与 Cedars-Sinai 的 QPS 软件的结果进行比较。
在模拟研究中，真实 EXT 与 TI（斜率=1.08；偏移量=-0.40%；r=0.99）和 Ts（斜率=0.90；偏移量=0.27%；r=0.99）方法具有很好的相关性，并且它们之间没有显著差异。同样，对患者研究中从 QPS 和模板方法获得的 EXT 也进行了很强的相关性测量（斜率=0.91；偏移量=0.45%；r=0.98）。使用模拟、物理体模和患者数据的 Ts 方法的平均扩展误差分别为 2.7% +/- 2.4%、0.9% +/- 0.5%、2.0% +/- 2.7%。
作者引入了一种用于 SPECT MPI 的半定量方法，该方法提供了一种针对心脏内灌注缺陷区域的患者特异性方法，而不是采用患者平均 NPD 方法。
