Khalilzadeh O, Sabel B, Sung Y, Parikh A, Phan C M, Dinkel J, Yoo A J, Romero J, Gupta R
Department of Radiology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA.
Clin Neuroradiol. 2014 Dec;24(4):365-71. doi: 10.1007/s00062-013-0268-0. Epub 2014 Jan 30.
CT hyperattenuation arising from iodinated contrast has a different temporal evolution than that arising due to hemorrhage. This paper presents a method for optimal discrimination between hemorrhage and iodinated contrast in a postintervention CT in stroke patients.
We analyzed the brain computed tomography (CT) scans of consecutive patients with intraparenchymal hyperattenuation due to hemorrhage (n=41), those due to iodinated contrast alone (n=24), and those due to contrast mixed with hemorrhage after reperfusion therapy (n=14) in stroke patients. The difference between the maximum enhancement in hyperattenuation in the affected area and the corresponding contralateral area, dubbed Relative Maximum Enhancement (RME), was tracked over time. We fitted regression models to the RME changes due to hemorrhage and contrast to describe their temporal decay, and then derived the optimal discriminant curve that distinguishes the two. A computer algorithm coregistered the baseline and follow-up CT scans and performed pixel-by-pixel comparison to determine hemorrhage and iodinated contrast based on the RME changes with respect to the discriminant curve.
For both hemorrhage (k= -0.004, R (2) =0.7) and iodinated contrast (k= -0.064, R (2) =0.9), the temporal evolution of RMEs were best fitted by exponential decay curves, with respective half-lives of 192.3 and 10.7 h. An exponential decay model (k= -0.026) for optimal discrimination of hemorrhage vs. contrast was fitted. The computer algorithm implementing this model was successful in predicting the presence of hemorrhage in a hyperdense lesion with sensitivity =93% and specificity =91%.
Intraparenchymal hemorrhage and contrast have markedly different decay half-lives that can be used to assess hemorrhage in a hyperdense lesion on a CT scan after intra-arterial therapy.
碘化造影剂引起的CT高密度影与出血引起的CT高密度影具有不同的时间演变过程。本文提出了一种在中风患者干预后CT中最佳区分出血和碘化造影剂的方法。
我们分析了连续的中风患者的脑部计算机断层扫描（CT），这些患者因出血导致脑实质内高密度影（n = 41），因单纯碘化造影剂导致脑实质内高密度影（n = 24），以及因再灌注治疗后造影剂与出血混合导致脑实质内高密度影（n = 14）。跟踪患侧区域与相应对侧区域高密度影最大强化之间的差异，即相对最大强化（RME）随时间的变化。我们对出血和造影剂导致的RME变化拟合回归模型以描述其时间衰减，然后得出区分两者的最佳判别曲线。一种计算机算法对基线和随访CT扫描进行配准，并基于相对于判别曲线的RME变化逐像素比较以确定出血和碘化造影剂。
对于出血（k = -0.004，R² = 0.7）和碘化造影剂（k = -0.064，R² = 0.9），RME的时间演变均最适合用指数衰减曲线拟合，各自的半衰期分别为192.3小时和10.7小时。拟合了用于最佳区分出血与造影剂的指数衰减模型（k = -0.026）。实施该模型的计算机算法成功预测了高密度病变中出血的存在，灵敏度 = 93%，特异性 = 91%。
脑实质内出血和造影剂具有明显不同的衰减半衰期，可用于评估动脉内治疗后CT扫描上高密度病变中的出血情况。
