Article Text


Dipoles and dementia
  1. C M Epstein
  1. Emory Clinic, Department of Neurology, 1365 Clifton Road, Atlanta, GA 30322, USA
  1. Correspondence to:
    Dr C M Epstein;

Statistics from

Can the link between MEG dipoles and cortical activity lead to increased applications for the unique capabilities of MEG?

For almost two decades magnetoencephalography (MEG) has been the stepchild of functional neuroimaging. Despite its evolution from single sensors to arrays of more than 100, with unmatched temporal resolution, the legitimacy of MEG as a clinical and research tool has been repeatedly challenged. On the one hand, neurophysiologists persistently—and appropriately—question the advantage of MEG over scalp-recorded electroencephalograms and evoked potentials when the latter are analysed by sophisticated algorithms. On the other hand, advocates of functional magnetic resonance imaging (MRI) and positron emission tomography have disparaged MEG because of the assumptions required to obtain unique solutions from magnetic field data. The paper by Maestú et al (this issue, page 208-212)1 provides an example of work that is establishing the capabilities of MEG.

Maestú et al used MEG to compare patients with Alzheimer’s disease and elderly subjects. They had previously found that reduced MEG activation was present in left temporal and parietal areas during a probe letter working memory task; this reduction correlated with results from neuropsychological assessments and activities of daily living. In the present study, they report that magnetic field measures correlate significantly with the relative volume of lateral and mesial temporal regions. Consistent with the long latency of cognitive evoked potentials, the MEG features of interest occurred more than 400 ms after stimulus onset. After a Bonferroni correction, the only correlation coefficient considered significant was that between left temporal lobe activity and the relative volume of the left hippocampus. The smaller the number of late MEG activity sources in the left temporal region, the greater the atrophy in the mesial portion of the left temporal lobe. Definable changes in MEG may correspond to the exquisite anatomic resolution of MRI.

Different types of algorithms, embodying different assumptions, have been employed to solve for the four-dimensional origin of electrical activity within the brain. The commonest algorithm models magnetic flux as the product of single equivalent current dipoles. Despite its simplicity, this approach has given good results when compared with invasive recordings. A straightforward MEG technique is to plow mathematically through a second or so of evoked potentials, solving for dipoles every 4 ms and identifying those that show high goodness of fit and a restricted intra cranial volume. Analysis of a single evoked potential can produce hundreds of dipoles scattered across multiple brain regions. This approach provides interesting results, and the assumptions behind it seem reasonable: but it’s fair to ask whether all those dipoles represent something “real”, or are merely artifacts of elaborate computer processing.

The report of Maestú and colleagues suggests that the plethora of dipoles is a valid reflection of brain activity. Significantly fewer of them are found in the atrophic left temporal lobes of Alzheimer’s disease patients compared to elderly subjects. Although we still don’t know what the individual dipoles mean, we have increasing reason to believe that they do reflect neuronal processing. We can hope that a greater understanding of the link between MEG dipoles and cortical activity will lead to increased applications for the unique capabilities of MEG.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles