The mind boggles.

Here's the article:

PHYSICS NEWS UPDATE The American Institute of Physics Bulletin of Physics News Number 680 April 8, 2004 by Phillip F. Schewe, Ben Stein

MRI WITH 80-NM RESOLUTION, far better than for the best medical scans, has been achieved with a device that combines atomic force microscope (AFM) and nuclear magnetic resonance (NMR; also known as magnetic resonance imaging, or MRI) technology. In the hybrid methodology called magnetic resonance force microscopy (MRFM), a tiny magnetized particle is attached to a cantilever which is then brought near a sample which surrounded by a coil that emits radio waves. When a tiny magnetic domain in the sample feels just the right amount of magnetic field from the nearby coil and magnetic particle it will vigorously interact with them resonantly. (The tiny volume being probed is referred to as a voxel, and the sample-coil-particle combination is equivalent to the setup in a standard MRI machine for imaging, say, a tumor.) The sample-particle resonant interaction causes the cantilever to oscillate (the particle on the cantilever is like a man bouncing resonantly, higher and higher, on a diving board). The oscillating cantilever, monitored with a laser beam, is then scanned from place to place, filling out a two-dimensional and then a three-dimensional map of the resonant interaction. (The scanned, oscillating cantilever plus laser readout is the AFM part of the setup.) The goal is not to help surgeons (the best medical MRI has a spatial resolution of about a tenth of a millimeter) but to be able to scan and image small objects---especially particles of biological importance, such as viruses and proteins---with atomic-scale resolution. In other words, you would like to increase the sensitivity so as to map the presence of single spins. The voxel in this case would be shrunk to less that than 1 nm. A new experiment at the University of Washington is far from reaching this goal, but researchers have improved sensitivity by a factor of almost 10,000 from the time of the earliest MRFM imaging papers in 1996. (For a report from 1997, see http://www.aip.org/enews/physnews/1997/split/pnu313-1.htm). The higher sensitivity in general comes by shrink the apparatus and cooling things (currently, to 80 K) as much as possible, the better to read out the oscillations and position the sample with greater accuracy. The Washington voxel of 80 nm---how big is it? One of the team members, John Sidles (206-543-3690, s idles@u.washington.edu) says that about a million of these voxels could fit inside a typical blood cell. (Chao, Dougherty, Garbini, Sidles, Review of Scientific Instruments, May 2004; website, courses.washington.edu/goodall/MRFM ) Other groups are working in this area and are attempting to marshal the requisite equipment needed for single-spin imaging. According to Joseph Shih-hui Chao, one of the authors, this would include millikelvin temperatures, 30-nm-sized magnetic particles, sub-nm positioning accuracy, and yet softer cantilevers.