Lyden P D, Zweifler R, Mahdavi Z, Lonzo L
Department of Neurosciences, University of California at San Diego, School of Medicine.
Stroke. 1994 Dec;25(12):2421-8. doi: 10.1161/01.str.25.12.2421.
Clinical stroke trials require objective and reproducible end point variables. Morphometry of cerebral structures, including infarct volume, provides numerical measures that represent the amount of tissue damaged and potentially salvaged by therapy. However, morphometry may be time-consuming and labor-intensive, and it requires standardization across multiple centers, which may be difficult to achieve in large multicenter trials. We developed a brain morphometry method that is unbiased, rapid, reliable, and based on well-accepted stereological techniques. We now extend this method to analysis of routine computed tomographic (CT) scans such as might be obtained during a clinical stroke trial.
We studied CT scans from 18 stroke patients and 14 asymptomatic control patients obtained over 5 years at the San Diego Veterans Administration Medical Center. Three observers independently measured the volume of the cranial vault, cerebrum, cortex, white matter, deep gray structures, ventricle, sulcal cerebrospinal fluid space, visible infarction, and cerebellum/brain stem.
The two patient groups were well matched demographically. The intracranial volume of 1400 +/- 40 mL in control subjects was not different from the 1311 +/- 41 mL in patients. Cerebral volume was 1250 +/- 36 mL compared with 1070 +/- 36 mL (control subjects versus patients, P < .001), and infarction volume was 55 +/- 16 mL in patients. For all structures, intraclass correlation coefficients among the observers ranged from 0.87 to 0.03; the best agreement was found for lesion, ventricle, and intracranial volume. White matter and cortex volume predicted the National Institutes of Health Stroke Scale score but not the late outcome scores on the Barthel Index or Rankin Scale. Each scan required 70 to 90 minutes for analysis.
We developed a stereological method for cerebral morphometry from CT scans that is reliable, rapid, and simple. The measurements are unbiased, can be made on slices of any known thickness, and are independent of machine variables. Our results are remarkably similar to values obtained with more labor-intensive methods. This method should be of use in large-scale, multicenter trials of stroke therapy.
临床中风试验需要客观且可重复的终点变量。包括梗死体积在内的脑结构形态测量提供了数值指标，可代表受损组织的数量以及可能通过治疗挽救的组织量。然而，形态测量可能耗时且费力，并且需要在多个中心进行标准化，这在大型多中心试验中可能难以实现。我们开发了一种基于公认的体视学技术的无偏倚、快速、可靠的脑形态测量方法。我们现在将此方法扩展到对常规计算机断层扫描（CT）图像的分析，例如在临床中风试验期间可能获取的图像。
我们研究了圣地亚哥退伍军人管理局医疗中心在5年期间获取的18例中风患者和14例无症状对照患者的CT图像。三名观察者独立测量颅腔、大脑、皮质、白质、深部灰质结构、脑室、脑沟脑脊液间隙、可见梗死灶以及小脑/脑干的体积。
两组患者在人口统计学特征上匹配良好。对照组的颅内体积为1400±40 mL，与患者组的1311±41 mL无差异。大脑体积对照组为1250±36 mL，患者组为1070±36 mL（对照组与患者组相比，P <.001），患者的梗死体积为55±16 mL。对于所有结构，观察者之间的组内相关系数范围为0.87至0.03；在病变、脑室和颅内体积方面一致性最佳。白质和皮质体积可预测美国国立卫生研究院卒中量表评分，但不能预测巴氏指数或Rankin量表的晚期结局评分。每次扫描分析需要70至90分钟。
我们开发了一种从CT图像进行脑形态测量的体视学方法，该方法可靠、快速且简单。测量无偏倚，可在任何已知厚度的切片上进行，且与机器变量无关。我们的结果与采用更费力方法获得的值非常相似。该方法应可用于中风治疗的大规模多中心试验。
