5 February 2002. Magnetic resonance imaging (MRI) of the brain shows great promise as a diagnostic tool. Neurodegenerative and psychiatric disorders can lead to changes in the brain that this imaging technique can pick up. These subtle differences are not just physiological, but can be due to alterations in the volume, shape, thickness or folding pattern of a particular section of brain tissue. Unfortunately, producing high-resolution MRI maps of such structural changes is a skilled and time-consuming process, which involves manually labeling each sectional image to generate a visual map.
In the 31 January Neuron, researchers led by Anders Dale, Massachusetts General Hospital, report on a new automated method that can reliably label the human brain in as little as 30 minutes. This mathematical mapping, built on a brain atlas generated manually, considers the probable accuracy of each label in relation to its neighbors before assigning its position.
The researchers found no statistical differences in volumes recorded by manual and automated labeling of fifteen distinct brain structures, from the lateral ventricles to the brainstem. Furthermore, when tested on 134 patients, the technique was capable of revealing statistical differences between the volumes of brain structures in Alzheimer's patients versus controls. Automated imaging showed that the former, for example, had dilated right and left lateral ventricles and reduced amygdala volumes, characteristics that have previously been noted in these patients.-Tom Fagan.
Reference:Fischl B, Salat DH, Busa E, Albert M, Dieterich M, Haselgrove C, van der Kouwe A, Killiany R, Kennedy D, Klaveness S, Montillo A, Makris N, Rosen B, Dale AM.
Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron 2002 January 31;(33):341-355. Abstract