Improved Estimation of Local Cerebral Glucose Metabolic Rate Using Bayes Regression Analysis of PET Scan Data

摘要

The current method for measurement of local cerebral metabolic rate of glucose (LCMRG) has relatively small errors when applied to normal healthy tissue. But when applied to ischemic regions in brains of stroke patients, the method gives estimates which average about 50% too low. Here we introduce a modified Bayes Regression (BR) to compute LCMRG, and use computer simulation studies to demonstrate that BR has relatively small errors for either ischemic or normal tissue.