Diffusion Tensor Magnetic Resonance Imaging of the Brain
Pierpaoli, Horkay, Pajevic, Barnett, Aldroubi, Barnet, Shrager, Cohen
We are continuing to develop Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) as a means to probe tissue microstructure and to assess and diagnose neurological and developmental disorders. DT-MRI measures a diffusion tensor of water on a pixel-by-pixel basis within tissue, both noninvasively and in vivo. It relates an effective diffusion tensor to the measured MR spin echo signal, estimating an effective diffusion tensor, D, in each pixel from a set of diffusion-weighted MR images and then calculating and displaying information derived from D, including local fiber-tract orientation, the mean-squared distance water molecules diffuse in any given direction, the orientationally averaged mean diffusivity, and other intrinsic scalar invariant quantities that are independent of the laboratory coordinate system. These scalar parameters behave like quantitative histological or physiological "stains," yet they are "developed" without requiring exogenous contrast agents or dyes.

FIGURE 30

For example, the orientationally averaged diffusivity (or Trace[D]) has been the most successful imaging parameter proposed to date to identify ischemic tissues during and following an acute stroke. Moreover, we have shown that DT-MRI is effective in identifying white matter degeneration (Wallerian degeneration) associated with chronic stroke. Studies with kittens have also shown that DT-MRI is useful in following early developmental changes occurring in cortical gray and white matter. Such changes cannot be detected with other imaging methods. A method to encode nerve fiber orientation in the brain using color, developed by Sinisa Pajevic and Carlo Pierpaoli, has allowed us to identify the main association, projection, and commissural white matter pathways in the human brain and even differentiate anatomical white matter pathways that have similar structure and composition but different spatial orientations. It has also allowed us to perform detailed studies of the brain's structural anatomy; previously, such studies could be performed only by using laborious, invasive histological methods. To assess anatomical connectivity between different functional regions in the brain, we have recently proposed and demonstrated a way to use DT-MRI data to trace out nerve fiber tract trajectories, which we call "fiber tractography." In this application, we compute the trajectory of a nerve fiber tract by continuously following the direction along which the apparent diffusivity is a maximum.

FIGURE 31

The development of DT-MRI also requires the development of new mathematical, statistical, and image processing concepts and constructs for analyzing the multidimensional data produced by this imaging method. Akram Aldroubi and Sinisa Pajevic have developed a general mathematical framework for obtaining a continuous, smooth approximation to the discrete, noisy, diffusion tensor field data that we obtain. This framework allows us to reduce the noise in our data and enables us to follow fibers more reliably. We have also derived the form of the parametric distribution governing the statistical variability of diffusion tensor data and have developed nonparametric (bootstrap) methods for determining features of their statistical distribution from experimental DT-MRI data. These developments are allowing us to apply powerful statistical hypothesis tests to address a wide variety of important biological and clinical questions that previously could be tackled only by using ad hoc methods. Finally, we are developing novel image processing methods to enable us to perform quantitative longitudinal or multicenter DT-MRI studies. These include methods to warp and register multidimensional images. To ensure that the DT-MRI acquisitions obtained with different scanners and at different sites are quantitative and of high quality, we are developing new polymeric phantoms with which we can calibrate DT-MRI measurement systems. Collectively, these developments are enhancing the utility and broadening the scope of applications of DT-MRI in medicine and biology.

Physical-Chemical Aspects of Cell and Tissue Excitability
Tasaki, Horkay, Basser
Excitability of cells and tissues is an essential physiological function that allows organisms to sense their environment and respond to it. The primary goal of our work is to explain key physical-chemical features of cell and tissue excitability, many aspects of which are still poorly understood. Widely accepted theories of nerve excitability fail to explain several anomalous phenomena that we have both observed and have shown are necessary for excitation to occur. These include volume and temperature changes of the superficial protoplasmic layer of nerve axons, which coincide with the action potential waveform. We have obtained further evidence that these changes accompany a phase transition that occurs in nerve cells, fibers, and synapses caused by the exchange of divalent cations such as calcium with monovalent cations such as sodium and potassium. Our previous experiments with perfused axons clearly implicate divalent/monovalent cation exchange as a mechanism by which nerve fibers can be excited in an "all or none" manner. To understand the physical-chemical basis of these temperature and volumetric changes, particularly how divalent/monovalent cation exchange can induce such changes in biomolecular assemblies, we are studying these processes in synthetic "biomimetic" anionic polymer gels under nearly physiological solution conditions. An advantage of studying the behavior of these gel model systems is that their structure, composition, and the interactions among their components can be carefully controlled, unlike in the case of living tissue. In particular, in synthetic polyacrylate gels, Ferenc Horkay has observed that minute changes in the concentration of divalent cations in the surrounding liquid can induce significant changes in chain stiffness in the gel, even if ion binding is weak and completely reversible. Various physical chemical and polymer physics-based techniques, including neutron scattering, osmotic swelling, and mechanical loading, provide complementary information with which to study these biologically relevant phenomena over a wide range of length scales. These basic studies are leading to a deeper understanding of the physical mechanisms underlying nerve excitation.

Functional Properties of Extracellular Matrix
Horkay, Basser, Potter, Amis, Washburn
The collagen network plays a critical role in determining functional properties of cartilage and other extracellular matrices. The collagen network exerts a retractive stress on the osmotically active proteoglycans that are trapped within it in much the same way a balloon's elastic membrane exerts a hydrostatic pressure on the gas contained within it. Until now, however, it has not been possible to measure the retractive stress of the collagen network independently of other constituents within the extracellular matrix. Recently, we devised a new methodology to determine this structural property of the collagen network. This new approach involves modeling the cartilage tissue matrix as a composite material consisting of two distinct phases: a collagen network and a proteoglycan (PG) solution trapped within it; applying various known levels of equilibrium osmotic stress; and using physical-chemical principles and independent experiments to determine useful "pressure-volume" relations for both the PG and collagen phases independently. In pilot studies, we used this approach to determine pressure-volume curves for the collagen network and the PG phases in native and in trypsin-treated normal human cartilage specimen as well as in cartilage specimens from osteoarthritic (OA) joints. In both normal and trypsin-treated specimens, collagen network stiffness appeared unchanged, whereas collagen network stiffness decreased in the OA specimen. Our findings highlight the role of the collagen network in limiting normal cartilage hydration and in ensuring a high PG concentration in the matrix, both of which are essential for effective load bearing in cartilage but are lost in OA. The data also suggest that the loss of collagen network stiffness, and not the loss or modification of PGs, may be the incipient event leading to the subsequent disintegraton of cartilage observed in OA.

As a visiting fellow in our section, Kimberlee Potter initiated microscopic Magnetic Resonance Imaging (MRI) studies designed to estimate, through noninvasive means, parameters of our mathematical model of cartilage swelling by attempting to relate the chemical composition of cartilage tissue grown in a hollow-fiber bioreactor to various measurable MRI quantities. Ferenc Horkay is now developing an instrument that will enable us to study swelling properties of extremely thin cartilage sections, permitting us to obtain a profile their functional properties with depth from the joint's articular surface.