BAYES REGRESSION ANALYSIS OF DEOXYGLUCOSE PET SCAN DATA

脱氧葡萄糖 PET 扫描数据的贝叶斯回归分析

• 批准号: 3404916
• 负责人: P DAVID WILSON
• 金额: $ 6.43万
• 依托单位: UNIVERSITY OF MARYLAND BALTIMORE
• 依托单位国家: 美国
• 财政年份: 1985
• 资助国家: 美国
• 起止时间: 1985-09-01 至 1987-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/3404916
• 关键词: brain metabolism; cerebral ischemia /hypoxia; computer simulation; glucose metabolism; mathematical model; nervous system disorder diagnosis; statistics /biometry; stroke

The overall aim of this project is to greatly improve the accuracy of the fluoro-deoxy-glucose (FDG) method of measurement of local cerebral metabolic rate of glucose (LCMRG) in ischemic brain tissue. Studies by Hawkins et al [6] have shown that when the existing FDG method is applied to ischemic tissue the resulting estimates and on the average, about 50% too low. The investigators have developed modifications of the data collection and analysis of the FDG method and performed statistical simulation studies to evaluate the relative accuracy of the new modified method, which is called Bayes Regression. These simulation studies indicate that the new modified method produces greatly improved relative accuracy in ischemic tisue: only 1 or 2% error. However this finding has not yet been validated in human stroke patients, and the main aim of this proposal is to obtain this validation. The secondary aims include the evaluation of other procedures proposed by others or by the investigators. The data used by Hawkins et al [6] mentioned above are available for analysis by Bayes Regression in this project. These data were obtained in the ischemic and contralateral normal regions of human stroke patients by repeated positron emission tomography(PET) scans for a period of 2 to 3 hours. The existing method used for routine clinical studies uses only a single PET scan taken at about 1 hour post-injection. While a data collection period of 2 to 3 hours is not practical for routine clinical studies, it does allow for computation of LCMRG by the "Direct" method which is known to produce accurate estimates of rate constants. A data collection period of 1 hour is practical for routine clinical studies; this collection period was used by Bayes Regression in the simulation studies, and only the first 1 hour's data from the stroke patients will be used in this project to validate the accuracy of Bayes Regression in ischemic tissue. The relative accuracy of LCMRG estimates by Bayes Regression will be assessed by comparison with estimates from the Direct method.

该项目的总体目标是大大提高 氟脱氧葡萄糖（FDG）法测量局部脑组织 缺血脑组织葡萄糖代谢率（LCMRG）。 的研究 Hawkins等人[6]已经表明，当应用现有的FDG方法时， 平均来说，大约有50%的 太低了 研究者对数据收集进行了修改， 分析FDG方法，并进行统计模拟研究， 评估新修改方法的相对准确度，该方法称为 贝叶斯回归 这些模拟研究表明， 方法在缺血组织中产生了大大提高的相对准确性：仅 1或2%的误差。 然而，这一发现尚未在人类中得到验证。 中风患者，这项建议的主要目的是获得这一点， 验证。 次要目的包括评价其他程序 由其他人或调查人员提出。 上述Hawkins等人[6]使用的数据可用于 贝叶斯回归分析在这个项目中。 这些数据是在 人中风患者的缺血和对侧正常区域， 重复正电子发射断层扫描（PET）扫描2 - 3个周期 小时 用于常规临床研究的现有方法仅使用 在注射后约1小时进行单次PET扫描。 而数据 2至3小时的收集期对于常规临床 研究，它允许通过“直接”方法计算LCMRG 已知其产生速率常数的精确估计。 数据 1小时的收集期对于常规临床研究是实用的;这 在模拟研究中使用贝叶斯回归的收集期， 只有中风患者的前1小时数据将用于 本项目旨在验证贝叶斯回归在缺血性脑卒中中的准确性， 组织. 贝叶斯回归估计的LCMRG的相对准确度将 通过与直接方法的估计值进行比较进行评估。