LeMay M, Stafford J L, Sandor T, Albert M, Haykal H, Zamani A
J Comput Assist Tomogr. 1986 Sep-Oct;10(5):802-9. doi: 10.1097/00004728-198609000-00018.
Three neuroradiologists perceptually evaluated CT of 24 patients with Alzheimer type dementia and 22 normal control subjects and made a dichotomous judgment for each case (i.e., normal control or Alzheimer type dementia). The mean percentage of patients correctly classified was 83.3%. The neuroradiologists also completed perceptual ratings on each scan. Thirteen regions were rated for atrophy on a scale of 0-4. Discriminant function analyses of several sets of perceptual atrophy ratings (optimized on an exploratory set and evaluated on a test set) showed that the perceptual ratings of temporal lobe regions produced an average accuracy of 88.57%. In contrast, only 74.26% of the cases were correctly classified when the discriminant functions were based on perceptual ratings of midventricular and supraventricular areas. Linear measures of atrophy correctly classified only 65.20% of the subjects. The results suggest that atrophy ratings of brain regions that show the characteristic macroscopic neuropathological changes of Alzheimer disease may be used by neuroradiologists to reach more accurate diagnostic decisions.
三位神经放射科医生对24例阿尔茨海默病型痴呆患者和22名正常对照者的CT进行了视觉评估，并对每个病例做出二分判断（即正常对照或阿尔茨海默病型痴呆）。正确分类的患者平均百分比为83.3%。神经放射科医生还对每次扫描完成了视觉评分。对13个区域的萎缩程度按0至4级进行评分。对几组视觉萎缩评分进行判别函数分析（在探索性数据集上进行优化，并在测试数据集上进行评估），结果显示颞叶区域的视觉评分平均准确率为88.57%。相比之下，当判别函数基于脑室中部和室上区域的视觉评分时，只有74.26%的病例被正确分类。萎缩的线性测量仅能正确分类65.20%的受试者。结果表明，神经放射科医生可以利用显示阿尔茨海默病特征性宏观神经病理变化的脑区萎缩评分来做出更准确的诊断决策。
