Magnetic resonance histology (MRH) is based on the same principles that underlie clinical magnetic resonance imaging (MRI). MRH differs from clinical MRI in several ways, but one difference is most fundamental. The resolution in MRI and MRH is very different. Though images of both methods are comprised of picture elements (pixels), each based on signal from a volume of tissue (voxel), the voxels are of very different scale.
A typical clinical MRI system encodes voxels at 1 x 1 mm in-plane, with a 3-10 mm slice thickness, so each voxel has a volume of 3-10 mm3. The human clinical MRI image (Figure 1a) is a 5 mm thick slice with in-plane resolution of 1 x 1 mm. The red area superimposed on the human brain shows the relative size of a mouse brain. When that area of the human scan is magnified (Figure 1b), graininess due to the large pixels is readily apparent. The MRH mouse brain image shown in (c) covers the same dimensions as the red area on the human brain (a), but is encoded at .035 x .035 x .035 mm. This much tinier pixel size demonstrates 100,000X higher spatial resolution than the clinical image, and shows details of the anatomy of the mouse brain.
Figure 1. Resolution: clinical MRI human brain image compared to MRH mouse brain image.
Since the voxels in an MRH image may be up to a million times smaller than those in an MRI image, the signal produced by a voxel will be up to a million times weaker. This poses substantial technical barriers to routine acquisition of MRH data. The Duke Center for In Vivo Microscopy is an NIH National Resource that was formed in 1983 to bring magnetic resonance microscopy from theory to practice. Over the last 18 years, the Duke Center has systematically addressed these challenges through the development of novel radio-frequency detectors, specialized encoding strategies, the use of very high field magnets, and the use of advanced visualization techniques.
Magnetic resonance microscopy has become a robust tool for the basic scientist. The ability to perform live animal longitudinal studies has been applied in drug discovery [1], toxicology [2], pulmonary disease [3], developmental biology [4], teratology [5]. MRH is a variation of MR microscopy which focuses on fixed specimens and is a more recent development.
Techniques at the Duke Center now allow routine acquisition of 3D volumetric MRH images that promise to revolutionize many of the fields where conventional optical histology is the current state of the art. MRPath was formed in 2000 to exploit these technologies to make MRH available to the broad community of basic and applied scientists. MRH provides a perfect compliment to other conventional optical imaging methods.
MRH is distinct from optical imaging in four ways :
|
The nondestructive nature of MRH is seen in Figure 2 which shows images from a fixative perfused C57BL/6J mouse. Since the "slices" are obtained by use of magnetic field gradients and radio-frequency pulses the specimen is never physically sectioned. This makes it is possible to examine the specimen in more than one cutting plane and at various resolutions. This example shows a coronal image with perfectly registered axial images.
Figure 2. MRH is nondestructive.
One of the strengths of both MRI and MRH is the unique soft tissue contrast available. The intrinsic contrast in the MRI or MRH image is due to water and how it is bound in the tissue. By altering acquisition parameters, different soft tissue structures can be highlighted, for example the density of water protons, the presence of fat, or the local diffusion of the water protons. There are many different types of contrast in MRI and MRH [6]. We refer to them collectively as "proton stains" by analogy to the myriad of chemical stains available for conventional optical microscopy.
Many of these "proton stains" have been shown to be exquisitely sensitive to subtle shifts in pathologic tissues. For example, in Figure 3 we see an MRH image from one hemisphere of a rat brain in which there is a neurologic lesion caused be exposure to a neurotoxin [7]. The MRH image on the right is a single slice from a whole intact brain. The conventional histology image on the right was generated by sectioning the specimen, dehydrating the tissues, and applying traditional chemical stains (Nissl). The subtle changes in water in the tissue accompanying the neurotoxic response produces a significant change in the MR properties rendering the lesion clearly visible.
Figure 3. MRH tissue staining
compared to conventional histology.
Scientists at the Duke Center for In Vivo Microscopy and MRPath have pioneered the concept of isotropic imaging [8]. Isotrophic image sets consist of a three dimensional array of voxels. The voxels are cubes and have the same extent in three dimensions. Most clinical MRI images are "slices" which have been individually excited and encoded. In the clinical image, the slice thickness is usually greater than the dimension of the pixels in the image plane.
Figure 4. Clinical imaging produces thick slices
made up of elongated voxels.
In Figure 4, the clinical human brain image is captured with 1 x 1 x 5 mm voxels. The coronal view (4a) is the "high resolution" plane, i.e. the plane with 1 x 1 mm resolution. The axial view (4b) projects the data the perpendicular plane with 1 x 5 mm resolution, showing the loss of resolution due to the thick slice.
Figure 5. MRH produces cubic voxels.
In contrast, the isotropic mouse brain image in Figure 5 (a) and (b) (at 35 x 35 x 35 micron resolution) shows identical resolution in both planes shown. The isotropic 3D nature of data produced at MRPath is a unique attribute of MRH technology. Since the results can be sliced in any chosen plane after imaging, the 3D nature aids the identification of complex structures. For example, in Figure 6a an oblique cutting plane through the mouse brain data set selects cerebellum and ventricles.
Figure 6. Cubic voxels permit cuts through
MRH data in any direction.
This brief introduction to MRH is not a full discussion of its potential. Examples in the Image Gallery demonstrate further opportunities for the application of MR histology.