Germano G, Erel J, Lewin H, Kavanagh P B, Berman D S
Department of Medicine, Cedars-Sinai Research Institute, Cedars-Sinai Medical Center, University of California Los Angeles School of Medicine, 90048, USA.
J Am Coll Cardiol. 1997 Nov 1;30(5):1360-7. doi: 10.1016/s0735-1097(97)00276-3.
We developed an automatic quantitative algorithm for the measurement of regional myocardial wall motion and wall thickening from three-dimensional gated technetium-99m sestamibi myocardial perfusion single-photon emission computed tomographic images.
The algorithm measures the motion of the three-dimensional endocardial surface using a modification of the centerline method, as well as wall thickening using both geometry (gaussian fit) and partial volume (counts).
The algorithm was tested using a "variable thickness" heart phantom, and the quantitative results were compared with visual segmental assessment of myocardial motion and thickening in 79 clinical patients with a wide range of ejection fractions (6% to 87%).
Phantom measurements of simulated motion and thickening were accurate regardless of the camera used (dual or triple detector), the angular span of reconstructed data (180 degrees or 360 degrees), the amount of motion (3 or 6 mm) or the amount of thickening (33%, 50% or 100%). Quantitative measurements of segmental motion and thickening in the patients were correlated with visual scores (r = 0.668, exact agreement 72.6%, kappa 0.433 and r = 0.550, exact agreement 74.7%, kappa 0.408, respectively). Significant inverse linear relations exist between the global (summed) visual motion score and the average quantitative motion, and between the global (summed) visual thickening score and the average quantitative thickening. Automatic quantitative ejection fraction measurements correlated extremely well with average quantitative motion (r = 0.929) and thickening (r = 0.959).
Our algorithm is accurate and may be the first automatic technique for the quantitative three-dimensional assessment of regional ventricular function in cardiology.
我们开发了一种自动定量算法，用于从三维门控锝-99m 甲氧基异丁基异腈心肌灌注单光子发射计算机断层扫描图像测量局部心肌壁运动和壁增厚。
该算法使用中心线方法的改进版本测量三维心内膜表面的运动，并使用几何方法（高斯拟合）和部分容积方法（计数）测量壁增厚。
使用“可变厚度”心脏模型对该算法进行测试，并将定量结果与 79 例射血分数范围广泛（6%至 87%）的临床患者的心肌运动和增厚的视觉节段评估进行比较。
无论使用何种相机（双探测器或三探测器）、重建数据的角度跨度（180 度或 360 度）、运动量（3 或 6 毫米）或增厚量（33%、50%或 100%），模拟运动和增厚的模型测量结果都是准确的。患者节段运动和增厚的定量测量与视觉评分相关（分别为 r = 0.668，完全一致率 72.6%，kappa 值 0.433 和 r = 0.550，完全一致率 74.7%，kappa 值 0.408）。整体（总和）视觉运动评分与平均定量运动之间，以及整体（总和）视觉增厚评分与平均定量增厚之间存在显著的负线性关系。自动定量射血分数测量与平均定量运动（r = 0.929）和增厚（r = 0.959）高度相关。
我们的算法准确，可能是心脏病学中用于局部心室功能定量三维评估的首个自动技术。
