Back to Search Start Over

MRI time series modeling of MS lesion development

Authors :
Dominik S. Meier
Charles R.G. Guttmann
Source :
NeuroImage. 32(2)
Publication Year :
2005

Abstract

A mathematical model was applied to new lesion formation in multiple sclerosis, as apparent on frequent T2-weighted MRI. The pathophysiologically motivated two-process model comprises two opposing nonlinear self-limiting processes, intended to represent degenerative and reparatory processes, respectively, investigating T2 activity from a dynamic/temporal rather than a spatial/static perspective. Parametric maps were obtained from the model to characterize the MRI dynamics of lesion development, answering the questions of how long new T2 lesion activity persists, how much residual damage/hyperintensity remains and how the T2 dynamics compare to those of contrast-enhancing MRI indicating active inflammation. 997 MRI examinations were analyzed, acquired weekly to monthly from 45 patients over a 1-year period. The model was applied to all pixels within 332 new lesions, capturing the time profiles with excellent fidelity (r = 0.89 ± 0.03 average correlation between model and image data). From this modeling perspective, the observed dynamics in new T2 lesions are in agreement with two opposing processes of longitudinal intensity change, such as inflammation and degeneration versus resorbtion and repair. On average, about one third of a new lesion consisted of transient signal change with little or no residual hyperintensity and activity of 10 weeks or less. Global lesion burden as MRI surrogate of disease activity may therefore be confounded by large amounts of transient hyperintensity. T2 activity also persisted significantly beyond the period of contrast enhancement, thereby defining MRI sensitivity toward a subacute phase of lesion development beyond blood–brain barrier patency. Concentric patterns of dynamic properties within a lesion were observed, consistent with concentric histological appearance of resulting MS plaques.

Details

ISSN :
10538119
Volume :
32
Issue :
2
Database :
OpenAIRE
Journal :
NeuroImage
Accession number :
edsair.doi.dedup.....43b029e02e6230d4a849fae50c1b4d8c