Researchers at Duke University have recently invented a technique for improving the spatial resolution of magnetic resonance imaging (MRI) by a factor of nearly 100,000x. Whereas routine clinical MRI scans contain 3-dimensional pixels (“voxels”) approximately 1mm x 1mm x 1mm, this new technique allows for voxels as small as 21.5 thousandths of a milimeter on each side.
This is fortunate for many mice, who in the future might no longer need to be sacrificed – but rather merely sedated – for precise neuroanatomical analysis. It is also fortunate for neuroscientists, as older histological staining methods were necessarily destructive and not necessarily completely accurate measurements of the living structures.
The new technique involves staining with a contrast agent to enhance the signal to noise ratio of MRI, with the advantageous side effect of reducing the amount of time for water molecules to relax after alignment by a strong magnetic field. This is the basis of the MR signal; tissues with less water show a different pattern of relaxation than those with more water, yielding an accurate measure of anatomical structure. Faster “relaxation times” therefore translates into faster image acquisition – great news for clinical applications where detailed anatomical scans might need to be taken quickly (and, as Michael Anes points out in the comments, perhaps even greater temporal resolution for fMRI).
There are also some fancy adaptations of the Fourier transform which are used to process the resulting magnetic signal. And we’re talking about massive magnetic fields here – on the order of 9 Tesla. (Some perspective: this is equivalent in power to the magnets to be used in the Large Hadron Collider at CERN.)