Time averaging and fitting of nonlinear metabolic changes: The issue of the time index choice applied to P-31 MRS investigation of muscle energetics

We present an exact analytical method dedicated to fitting time-dependent e xponential-like changes in MR spectra. As an illustration, this method has been applied to fitting metabolic changes recorded by P-31 MRS in human ske letal muscle occurring during a rest-exercise-recovery protocol. When recor ding metabolic changes with the accumulative method, the time averaging of the MR signals implies the choice of a time index for fitting any changes i n the features of the associated MR spectra. A critical examination of the different ways (constant, linear, and exponential) of choosing the time ind ex is reported. By numerical analysis, we have calculated the errors genera ted by the three methods and we have compared their sensitivity to noise. I n the case of skeletal muscle, both constant and linear methods introduce l arge and uncontrolled errors for the whole set of metabolic parameters deri ved from [PCr] changes. In contrast, the exponential method affords a relia ble estimation of critical parameters in muscle bioenergetics in both norma l and pathological situations. This method is very easy to implement and pr ovides an exact analytical solution to fitting changes in MR spectra record ed by the accumulative method. (C) 2001 Academic Press.