Sequential assessments of cerebral metabolic rates for glucose: Issues related to compartmental analysis

Date of Award




Degree Name

Doctor of Philosophy (Ph.D.)


Biomedical Engineering

First Committee Member

Myron D. Ginsberg, Committee Chair


Positron emission tomography (PET) and autoradiography (AR) have been used to study the brain function using radiolabeled deoxyglucose (DG). It can be advantageous to perform a sequential resting-activation paradigm on individual subjects. Deducing facts from an activation study conducted right after a resting study relies on separating observed radiation measurements into two parts, those due to accumulation of tracer during the second study and those connects with accumulation of tracer during the first study. Because a physical separation is impossible, a mathematical separation is used. The core of this work is to establish a theoretical and conceptual basis for the use of compartmental systems in the mathematical separation technique. An accurate separation not only relies on a plausible compartmental system but also on an accurate knowledge of the values of compartmental coefficients. Because the knowledge of coefficients and the complexity of the compartmental system are interconnected, the accuracy of separation relies on understanding how they are connected.Theoretical foundations for sequential studies are established successfully for a relatively simple three-compartmental four rate-constant system and a more elaborate four-compartmental five rate-constant system. The issues concerning the use of compartmental systems for sequential studies are examined in detail through computer simulations. The results of simulations demonstrated that not all the rate-constant values employed in the more elaborate compartmental system are estimated accurately and precisely. The two rate-constants ($k\sbsp{4}{*}$ and $k\sbsp{5}{*}$) involving the loss of DG-6-phosphate in tissue are estimated poorly. The relatively simple system, with fewer rate-constants to be resolved, provides more balanced measurements for the four rate-constants. The conventional wisdom in using population rate-constant values for mathematical separation, especially population values of $k\sbsp{4}{*}$ and $k\sbsp{5}{*}$, may not be valid. An important result from the simulations is that the mathematical separation can be performed with high degree of accuracy and precision even though individual processes are not resolved. Hence a potentially very useful alternative operational equation is proposed to measure the metabolic rate for glucose for sequential studies.


Engineering, Biomedical

Link to Full Text