Automated quantification of myocardial perfusion SPECT using simplified normal limits
Automated quantification of myocardial perfusion SPECT using simplified normal limits
复制标题
DOI：
10.1016/j.nuclcard.2004.10.006
复制
发表时间：
2005-01-01
影响因子：
2.4
通讯作者：
Germano, G
中科院分区：
文献类型：
--
作者：
Slomka, PJ;Nishina, H;Germano, G
背景为了简化心肌灌注SPECT（MPS）的正常限值的开发，我们实施了一个量化方案，其中正常限值是在不对异常扫描进行视觉评分或优化区域阈值的情况下得出的。正常限值来自男性（n = 40）和女性（n = 40）低可能性患者的同日TI-201静息/Tc-99-m-sestamibi负荷扫描。缺损程度、总灌注缺损（TPD）和局部灌注程度通过与极图坐标中的正常限值进行比较而得出。分析了256例连续患者的MPS扫描，这些患者没有已知的冠状动脉疾病，接受了冠状动脉造影术。新的量化方法（TPD）与我们以前开发的量化系统和视觉评分进行了比较。TPD检测50%或以上狭窄的受试者操作特征曲线下面积（0.88 +/- 0.02）高于视觉评分（0.83 +/- 0.03）（P = 0.039）或标准定量（0.82 +/- 0.03）（P = 0.004）。对于70%或以上狭窄的检测，TPD（0.89 +/- 0.02）高于标准定量（0.85 +/- 0.02）（P = 0.014）。TPD的敏感性和特异性分别为93%和79%;视觉评分分别为81%和85%;标准定量分别为80%和73%。使用应力模式特定的正常限值并没有提高性能。简化的量化实现了优于或等同于视觉评分或基于异常阈值的每段视觉优化的量化的性能。
Background. To simplify development of normal limits for myocardial perfusion SPECT (MPS), we implemented a quantification scheme in which normal limits are derived without visual scoring of abnormal scans or optimization of regional thresholds.Methods and Results. Normal limits were derived from same-day TI-201 rest/Tc-99-m-sestamibi stress scans of male (n = 40) and female (n = 40) low-likelihood patients. Defect extent, total perfusion deficit (TPD), and regional perfusion extents were derived by comparison to normal limits in polar-map coordinates. MPS scans from 256 consecutive patients without known coronary artery disease, who underwent coronary angiography, were analyzed. The new method of quantification (TPD) was compared with our previously developed quantification system and visual scoring. The receiver operator characteristic area under the curve for detection of 50% or greater stenoses by TPD (0.88 +/- 0.02) was higher than by visual scoring (0.83 +/- 0.03) (P = .039) or standard quantification (0.82 +/- 0.03) (P = .004). For detection of 70% or greater stenoses, it was higher for TPD (0.89 +/- 0.02) than for standard quantification (0.85 +/- 0.02) (P = .014). Sensitivity and specificity were 93% and 79%, respectively, for TPD; 81% and 85%, respectively, for visual scoring; and 80% and 73%, respectively, for standard quantification. The use of stress mode-specific normal limits did not improve performance.Conclusion. Simplified quantification achieves performance better than or equivalent to visual scoring or quantification based on per-segment visual optimization of abnormality thresholds.