Neuroradiology 2002 Jul;44(7):586-91
Wilson M, Morgan PS, Blumhardt LD.
University of Nottingham, Department of Neurology, Royal Preston Hospital, Preston, PR2 9HT, United Kingdom
Quantitative diffusion-weighted MRI has been applied to the study of neurological diseases, including multiple sclerosis, where the molecular self-diffusion coefficient D has been measured in both lesions and normal-appearing white matter.
Histograms of D have been used as a novel measure of the "lesion load", with potential applications that include the monitoring of efficacy in new treatment trials.
However different ways of measuring D may affect its value, making comparison between different centres and research groups impossible.
We aimed to assess the effect, if any, of using two different MRI sequences on the value of D.
We studied 13 healthy volunteers, using two different quantitative diffusion sequences (including different b(max) values and gradient applications).
Maps of D were analysed using both regions of interest (ROI) in white matter and "whole brain" histograms, and compared between the two sequences.
In addition, we studied three standardised test liquids (with known values of D) using both sequences.
Histograms from the two sequences had different distributions, with a greater spread and higher peak position from the sequence with lower b(max).
This greater spread of D was also evident in the white matter and test liquid ROI.
"Limits of agreement" analysis demonstrated that the differences could be clinically relevant, despite significant correlations between the sequences obtained using simple rank methods.
We conclude that different quantitative diffusion sequences are unlikely to produce directly comparable values of D, particularly if different b(max) values are used.
In addition, the use of inappropriate statistical tests may give false impressions of close agreement.
Standardisation of methods for the measurement of D are required if these techniques are to become useful tools, for example in monitoring changes in the disease burden of multiple sclerosis.