Application of continuous relaxation time distributions to the fitting of data from model systmes and excised tissue

Abstract

Biological systems exhibit heterogeneity at many different levels, leading to the expectation of multiple relaxation time components for water protons in tissue samples. Traditional methods which fit the relaxation data to an a priori number of discrete components are open to observer bias in their interpretation of this data, and moreover, are intuitively less realistic for heterogeneous systems than methods which produce continuous relaxation time distributions. Previous validations of continuous distribution techniques have been made on simulated data assuming uniform Gaussian noise. In the current work we have investigated the ability of one particular linear inverse theory technique to reproduce known relaxation time distributions from the data on a controllable model system. Furthermore, using the experience gained on the model system, we have applied this same technique to the analysis of in vitro relaxation time measurements on excised brain tissue and found for water protons in white matter, four reproducible components for the transverse relaxation, whereas gray matter gave rise to only two. The longitudinal relaxation displayed only one component in either white matter or gray matter.