WO2007092740A2 - Method and system for the amplification of nuclear magnetic resonance imaging - Google Patents

Abstract

A method and system to enhance MRI contrast based on nonlinear feedback fields induced by the spin of the imaged molecules are provided. It is shown that while contrast in magnetic resonance imaging (MRI) is often limited by the dependence of the magnetization on MR parameters that may not vary significantly throughout the sample, through use of the inventive system, changes in the distribution of an initially unstable magnetization configuration act back on the spins through the feedback fields to amplify contrast. Demonstrations of avalanching amplification of MRI contrast due to small differences in spin density or resonance frequency under the feedback interactions of the distant dipolar field and/or radiation damping in phantoms and in vitro human brain tissue are also provided. These demonstrations show up to 20 times improved contrast in epileptogenic lesions and malignant brain tumors, tissues with minimal contrast differences in routine MRI, which suggests that feedback-based contrast enhancement may lead to improved lesion characterization, among other potential biomedical applications.

Description

METHOD AND SYSTEM FOR THE AMPLIFICATION OF NUCLEAR MAGNETIC RESONANCE IMAGING

STATEMENT OF FEDERAL FUNDING The invention described herein was made in the performance of work under National

Science Foundation Grant No CHE-0349362, and the Federal Government has certain rights thereto.

FIELD OF THE INVENTION

The present invention is directed to a system and method for enhancing images from nuclear magnetic resonance devices, and more particularly to systems and methods for contrast enhancement of such images through nonlinear feedback.

BACKGROUND OF THE INVENTION

Nuclear magnetic resonance, or "NMR" as it is abbreviated by scientists, is a phenomenon that occurs when the nuclei of certain atoms are immersed in a static magnetic field and exposed to a second oscillating magnetic field. Some nuclei experience this phenomenon, and others do not, dependent upon whether they possess a property called spin. If the spin quantum number of the nucleus is non-zero, then it possesses spin angular momentum, which results in a magnetic dipole moment that can interact with the external magnetic field. Felix Bloch and Edward Purcell, both of whom were awarded the Nobel Prize in 1952, discovered the magnetic resonance phenomenon independently in 1946. In the period between 1950 and 1970, NMR was developed and used for the analysis of chemical and physical systems.

Since its discovery, researchers have found that NMR can be used to study physical, chemical, and biological properties of many different types of matter. As a consequence, NMR spectroscopy finds applications in many areas of science. For example, NMR spectroscopy is routinely used by chemists to study chemical structure using simple one- dimensional techniques. Two-dimensional techniques are used to determine the structure of more complicated molecules. These techniques complement and, in some cases, replace x-ray crystallography for the determination of protein structure, particularly in solution. In addition, time domain NMR spectroscopic techniques are used to probe molecular dynamics in solutions. Solid state NMR spectroscopy is used to determine the molecular structure of solids. Other scientists have developed NMR methods of measuring diffusion coefficients. The versatility of NMR makes it pervasive in the sciences.

One well-known NMR technique is magnetic resonance imaging "MRI", which is an imaging technique used primarily in medical settings to produce high quality images of soft tissues in the human body. In 1971 Raymond Damadian showed that the relaxation times of water protons in tissues and tumors differed, thus motivating scientists to consider magnetic resonance for the detection of disease. (See, e.g., R. V. Damadian, "Tumor Detection by Nuclear Magnetic Resonance," Science 171:1151 (March 19, 1971)). In 1973 the x-ray-based computerized tomography (CT) was introduced by Hounsfield. (G.N. Hounsfield, Br. J. Radiol. 46:1016-1022 (1973)). Later that same year, Paul Lauterbur demonstrated magnetic resonance-based imaging by applying magnetic field gradients to image the spatial distribution of water protons in objects, in combination with image reconstruction methods developed for CT. (P. C. Lauterbur, "Image Formation By Induced Local Interactions - Examples Employing Nuclear Magnetic-Resonance," Nature 242:190-191 (1973)). In 1975 Richard Ernst proposed magnetic resonance imaging using phase and frequency encoding and the Fourier Transform. (A. Kumar, D. Welti, R.R. Ernst, "NMR Fourier zeugmatography," J. Magn. Reson. 18:69-83 (1975).) This technique became the basis of current MRI techniques and was further modified for more widespread use by Peter Mansfield. (P. Mansfield, "Multi-planar Image-formation using NMR spin echoes," J. Phys. C 10: L55-L58 (1977)). Edelstein and coworkers demonstrated imaging of the body using Ernst's technique in 1980. A single image could be acquired in approximately five minutes by this technique. In 1992 functional MRI (fMRI) was developed. (See, e.g., K.K Kwong et al., "Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation," Proc. Natl. Acad. ScL (USA), 89:5675 (1992); and P.A. Bandettini et. al., "Time course EPI of human brain function during task activation," Magn. Reson. Med. 25:390-397 (1992).) Accordingly, although MRI started out as a tomographic imaging technique, that is it produced an image of the NMR signal in a thin slice through the human body, MRI has advanced beyond a tomographic imaging technique to a volume imaging technique. The disclosures of these background references are all incorporated herein by reference. However, despite all of these advancements, NMR and MRI are still limited because of the inability of current systems to generate sufficient contrast between the imaged elements. The ability to generate sufficient image contrast based on small variations in local magnetic resonance "MR" parameters is crucial for the noninvasive mapping of structure and function by MR microscopy and clinical MRI. The delineation of distinct internal structures becomes particularly challenging when the MR properties do not vary significantly, leading to imperceptible contrast changes. For example, conventional MR imaging methods are often unable to distinguish between diseased and neighboring normal tissues when the underlying pathology only causes a slight change in the corresponding MR parameters. (See, e.g., S. Y. Huang et al., "Avalanching amplification of MR imaging lesion contrast by nonlinear feedback," Science (2005), submitted; and S. Y. Huang et al., "Improving MRI differentiation of gray and white matter in epileptogenic lesions based on nonlinear feedback," Magn. Reson. Med. (2005), submitted, the disclosures of which are incorporated herein by reference.) MRI contrast arises from the dependence of the magnetization on the MR parameters, as prescribed by the equation of motion governing spin dynamics in liquids, the Bloch equations (See, e.g., "Magnetic resonance imaging: Physical principles and sequence design," New York: Wiley; 1999. pg. 914, the disclosure of which is incorporated herein by reference.) Pulse sequence parameters, such as pulse excitation profiles or variable delays, may be flexibly engineered to impart weighting on the magnetization m(r,t) for specified MR properties, most commonly the longitudinal and transverse relaxation times (Tj and T₂, respectively) and spin density. Limitations in contrast are frequently encountered in high- field MR imaging and microscopy, in which changes in T₁ and T₂ lead to convergence of relaxation-based contrast (J. M. Tyszka et al., "Magnetic resonance microscopy: recent advances and applications," Curr. Opin. Biotechnol. 16:93-99 (2005), the disclosure of which is incorporated herein by reference.) External, metal-ion based contrast agents, such as gadolinium-bound ligands (P. Caravan et al., "Gadolinium(III) chelates as MRI contrast agents: Structure, dynamics, and applications," Chem. Rev. 99:2293-2352 (1999), the disclosure of which is incorporated herein by reference) or ferromagnetic nanoparticles (H. J. Weinmann et al., "Tissue-specific MR contrast agents," Eur. J. Radiol. 46:33-44 (2003); and J. W. M. Bulte, D. L. Kraitchman, "Iron oxide MR contrast agents for molecular and cellular imaging," NMR Biomed. 17:484-499 (2004), the disclosures of which are disclosed herein by reference), may be used to enhance contrast due to MR relaxation properties. However, the relatively high concentration of contrast agents that must be delivered (E. T. Ahrens et al., A model for MRI contrast enhancement using T-I agents. Proc. Natl. Acad. Sci. USA 95:8443- 8448 (1998), the disclosure of which is disclosed herein by reference), combined with their toxicity and possibly poor penetration in vivo, e.g., due to the blood-brain boundary, sets an upper limit on the attainable contrast, which may prove insufficient for practical use. Accordingly, new methods and systems for enhancing the contrast provided by NMR and MR devices are needed.

SUMMARY OF THE INVENTION

The present invention is directed to methods and systems for NMR and MRI contrast enhancement using the intrinsic spin dynamics in the presence of nonlinear feedback interactions.

BRIEF DESCRIPTION OF THE DRAWINGS

Other objects and advantages of the invention will be evident to one of ordinary skill in the art from the following detailed description, made with reference to the accompanying drawings, in which:

FIG. 1 shows a schematic diagram of an exemplary contrast enhancement system in accordance with the current invention;

FIG. 2 provides data plots of the avalanching amplification of MRI contrast by the individual and joint feedback fields of the DDF (BJ) and radiation damping (B,), demonstrated numerically and experimentally in simple phantoms;

FIG. 3 shows a block diagram of an exemplary active feedback circuit for amplifying the radiation damping field at lower fields using conventional probes;

FIG 4 shows exemplary images of (A) radiation damping-enhanced (RD) and (B) T₂- weighted (T₂) MR images of human brain tissue excised from the left posterior parietal- occipital lobe of a pediatric patient with cortical dysplasia, compared with (C) histopathology and (D) gross anatomy;

FIG. 5 shows a comparison of (A) radiation damping-enhanced (RD) and conventional (B) T ₂-, (C) Tj-, and (D) proton density (PD) MR images, corresponding to those shown in Fig. 3; FIG. 6 shows a comparison of (A) histopathology with (B) radiation damping- enhanced and (C) conventional T₂-, (D) Ti-, and (E) proton density MR images of brain tissue excised from the left temporal lobe of an adult patient with glioblastoma multiforme;

FIG. 7 shows exemplary images of (A) histopathology, (B) radiation damping- enhanced, and (C) ^-weighted MR images of brain tissue taken from the same patient as in Fig. 5; FIG. 8 shows a comparison of (A) histopathology with (B) radiation damping- enhanced and (C) conventional T2-, (D) Ti-, and (E) proton density MR images, corresponding to the images shown in Fig. 6;

FIG. 9 shows exemplary images of (A) Radiation damping-enhanced MR image, (B) histopathology, (C) ^-weighted MR image, and (D) gross anatomy of another brain section taken from the same patient as in Figs. 5-7;

FIG. 10 shows a comparison of (A) radiation damping-enhanced and (B) conventional T ₂-, (C) T_J-, and (D) proton density MR images, corresponding to those shown in Fig. 8;

FIG. 11 shows graphical plots of data for the contrast-to-noise ratios for radiation damping-enhanced and conventional Ti -weighted, T₂-weighted, and proton density images;

FTG. 12 shows exemplary in vivo feedback-based contrast enhancement images in Poecilia reticulata (common guppy fish): simulated (A) radiation damping-enhanced and (B) feedback-enhanced images under radiation damping and the DDF, compared with experimental (C) radiation damping-enhanced and (D) joint feedback field-enhanced MR images at 14.1-T of an in vivo guppy fish placed in a 5-mm sample tube (sagittal image shown at left with axial cross-section marked);

FIG. 13 shows exemplary in vivo feedback-based contrast enhancement images in

Xenopus laevis (African) clawed frog) embryos: evolution of the magnetization under (A) radiation damping and (B) and (C) radiation damping combined with RF pulses produced images with improved contrast compared with representative (D) T₂ -weighted, (E) T_\- weighted, and (F) proton density MR images; and

FIG. 14 shows exemplary in vivo images of mice acquired by: (A) active feedback- enhanced imaging; (B) proton density imaging;. (C) Ty-weighted imaging; and (D) T₂*- weighted imaging.

DETAILED DESCRIPTION OF THE INVENTION

The current invention is directed to a system and method for MRI contrast enhancement that manipulates the intrinsic spin dynamics in the presence of nonlinear feedback interactions. This approach yields robust image contrast sensitive to small variations in versatile MR parameters that is not seen in conventional MR images. Discussion of Methodology

The basis of feedback-based contrast enhancement of the current invention is outlined in detail below. An exemplary embodiment of the method of the current invention is outlined in Scheme 1 and explained as follows: To understand the physical origin of contrast in MRI, consider the classical formulation of spin dynamics in liquids, known as the Bloch equations (FIG. 1, center). The normalized water ¹H magnetization in a voxel at position r at time t is defined by Equation 1: m(r,ϊ)≡M(r,t)/M₀ (Eq. 1)

(Mo is the equilibrium ¹H magnetization of pure water), evolves under the Bloch equations, where δύϊ≡Cϋ-aio is the resonance offset from a reference frame rotating at frequency Cϋo, B(r,t) represents the effective field, D is the diffusion coefficient, Tj and Ti are the longitudinal and transverse relaxation times, respectively, the equilibrium magnetization m_eq(r)≡M_eq(r)/Mo (normalized with respect to Mo), and {x, y,z} are the Cartesian basis vectors. As shown in FIG. 1, to enhance the dependence of the magnetization m(r,t) (10) on specific MR properties and hence improve MRI contrast, the current invention employs a local field B(r,t) (12) that explicitly depends on m(r,t) and renders the Bloch equations nonlinear. For solutions at high field with abundant high-gyromagnetic ratio (f) nuclei, e.g., ¹H, magnetization-dependent contributions to B(r,t) mainly come from two feedback fields: (1) the distant dipolar field (DDF) (DDF is further described in Deville G, Bernier M, Delrieux JM, "NMR multiple echoes observed in solid He-3," Phys. Rev. B 1979; 19:5666- 5688; and Warren WS, et. al., "Generation of impossible cross-peaks between bulk water and biomolecules in solution NMR," Science 1993;262:2005-2009, the disclosures of which are incorporated herein by reference), Ba(r,t), and (2) radiation damping (as described further in Bloembergen N, Pound RV, "Radiation damping in magnetic resonance experiments," Phys. Rev. 1954; 95:8-12; and Warren WS, Hammes SL, Bates JL, "Dynamics of radiation damping in nuclear magnetic resonance," J. Chem. Phys. 1989;91:5895-5904, the disclosures of which are incorporated herein by reference), B^t). A simple pulse sequence first prepares the magnetization in an initial unstable configuration m(r,to) (Jeener, J., "Dynamical instabilities in liquid nuclear magnetic resonance experiments with large nuclear magnetization, with and without pulse field gradients," J. Chem. Phys. 2002;l 16:8439-8446, the disclosure of which is incorporated herein by reference) under the feedback fields (also called reaction fields). Again, as shown in FIG. 1, subsequent evolution of m(r,to) (14) under the MR parameters in the Bloch equations generates small variations in the resulting magnetization distribution m(r,to+Δt) (16), which are reflected in the feedback field B(r,to+Δt)=B(m(r,to+Δt)). B(r,to+Δt) acts on m(r,to+Δt) to bring it away from the initial unstable state with ever-increasing efficiency, such that changes in the magnetization distribution act back on the magnetization through the feedback field to amplify contrast in a positive feedback cycle. The magnetization distribution is then imaged by a spatially encoding detection sequence. Contrast enhancement is triggered by the smallest changes in the magnetization distribution and builds up rapidly to reflect the underlying MR parameters, leading us to refer to such enhancement as "avalanching amplification."

The inventive method, as shown in the exemplary embodiment provided in FIG. 1 and the above description, can be adapted to enhance NMR contrast arising from a variety of MR properties. As a demonstration, Applicants have applied this approach to enhance contrast due to differences in spin density and resonance frequency. The specific mechanisms producing avalanching amplification of MRI contrast under the DDF, radiation damping, and the joint reaction fields are demonstrated in FTG. 2 through simulations (neglecting Tj, T₂, and diffusion processes for simplicity) and experiments on simple imaging phantoms. As proof of the principle, the feedback-based contrast enhancement is then demonstrated experimentally on in vitro unfixed human brain tissue samples excised from epileptogenic and cancerous lesions (FIGs. 3-9).

The first feedback field, the DDF, arises from long-range residual dipolar couplings that survive motional averaging in solution (as described in greater detail in W. S. Warren, et. al., Science 262, 2005 (1993), the disclosure of which is incorporated herein by reference.). In liquids, diffusion only averages out dipolar couplings between spins separated by distances less than the average diffusion length (-microns). The vector sum of the DDF from uniform magnetization in a spherical sample vanishes; however, this spherical symmetry can be broken by sample geometry or spatial modulation of the magnetization by gradients. The DDF B_d(r,t) is expressed as in Equation 2, below:

γB_d(τ,t) (SI units) (Eq. 2)

where cosθ_rr<=(z-z')/\r-r'\ and μo is the magnetic permeability of a vacuum. The characteristic DDF time constant T_d is approximately 65 ms for pure water at 14.1 T and 300 K. While B_d(r,t) is a global microscopic reaction field, as provided in Equation 3, below, for magnetization that is fully modulated along a single spatial direction s, B^r,t) can be well- approximated as a function of the local magnetization m(s,t), where s≡r s (as described in greater detail in W. S. Warren, S. Lee, W. Richter, S. Vathyam, Chem. Phys. Lett. 241, 207 (1995), the disclosure of which is incorporated herein by reference):

YB_d(s,t) (Eq. 3)

In short, the current invention applies the instability of m(r,t) under the DDF to enhance contrast between regions with small differences in spin density. (For a further discussion of instability under the DDF see J. Jeener, Phys. Rev. Lett. 82, 1772 (1999); and J. Jeener, /. Chem. Phys. 116, 8439 (2002), the disclosures of which are incorporated herein by reference.) In practice, however, contrast enhancement under the DDF is more apparent at longer evolution times (~s) (FIG. 2A), and is thus most applicable to samples with long T₂ relaxation times.

To amplify contrast between regions differing in resonance frequency (FIG. 2B), the current invention uses the second feedback field, radiation damping. Radiation damping is a macroscopic field that is fed back to the spins through the induced current in the receiver coil, as governed by Lenz's law. This reaction field creates a torque to rotate the bulk magnetization vector back to the +z-axis at a rate proportional to the magnitude of the net transverse magnetization. When the sample is on resonance in a perfectly tuned probe, the radiation damping field B_r(t) can be described by Equation 4, below.

where η is the coil filling factor and Q is the probe quality factor. For typical high-β probes (£)=500), the radiation damping time constant τ_τ is on the order of 10 ms. For superconducting probes or cryoprobes, Q values as large as 10,000 may be achieved (as discussed further in R. D. Black et al, Science 259, 793 (1993), the disclosure of which is incorporated herein by reference), reducing τ_r to 1 ms or less. Specifically, in FIG. 2 contrast enhancement under the radiation damping field B^t), is simulated by the evolution of a concentric cylindrical phantom containing solutions with slightly different resonance frequencies.

FIG. 2 compares the avalanching amplification of MRI contrast by the individual and joint feedback fields of the DDF (B_d) and radiation damping (B_r), which are demonstrated numerically and experimentally in simple phantoms. Phantoms are modeled by water ¹H in the outer region (spin density \m(r,t=0)\=l, resonance offset δω=0) and ¹H with a slight difference in MR properties in the inner region. (FIG. 2A) Simulations of contrast enhancement under the DDF only, with spin density difference δm in the inner region. Top: Contrast Δm_z, defined as the difference in m_z between representative voxels taken from the two regions, marked by "x" in the bottom figure. Bottom: Cross-sections of m_z averaged along the z-axis, taken along the center of the sample. (FIG. 2B) Simulations of contrast enhancement under radiation damping only, with resonance frequency difference δω. (FIG. 2C, left) Simulations of contrast enhancement under the joint reaction fields, with resonance frequency difference δω, using the pulse sequence shown in (FIG. 2A). (FIG. 2C, right) Experimental results showing contrast enhancement under the joint reaction fields for water in a 5 mm tube with a 1-mm inner capillary (off -centered) containing 5% ethanol solution by volume. Experimental pulse sequence is shown in (FIG. 2C), where Gr=5G-ms/cm for the first gradient (second gradient is a spoil gradient), and echo-planar imaging (FED mode) was used to image the sample (128x128 voxels). Dynamics were simulated by numerically integrating the nonlinear Bloch equations using the approach in reference (see, T. Enss, S. Ahn, W. S. Warren, Chem. Phys. Lett. 305, 101 (1999), the disclosure of which is incorporated herein by reference) without relaxation and diffusion effects ((GTj^SG-ms/cm, Tr = 10 ms, T_d = 69 ms). Phantom sample was 0.47³ mm³ sampled by 64x64x16 points. Diameter of inner region was 0.073 mm, sampled by 10 voxels across As discussed above, in FIG. 2A a concentric cylindrical phantom containing water with a slight difference in proton density, dm, between the inner and outer compartments is considered. In the DDF scheme (FIG. 2A), following the preparation sequence [O_x-(GT)₁], the transverse magnetization vectors m₊(r,t)≡m_x(r,t)+im_y(r,t) are modulated along the z-axis in a helical configuration. For samples with uniform proton density, following the z-gradient, m₊(r,t) is aligned with B_d,+(r,t)≡B_ci_yX(r,t)+iB_cι_ιy(r,t) for all voxels in the same xy plane, and the total magnetization precesses uniformly under B^ _z (see, e.g., Eq. 3). However, for the spatial distribution of proton densities considered here, m₊(r,t) in the inner and outer cylinders precess at slightly different rates under respective B^₁ fields whose difference is proportional to an. As m₊(r,t) in the two regions evolve out of phase, the phases of the corresponding B_d,+(r,t) fields also change accordingly. In addition, due to an, the phase angles between B_d,+(r,t) and m₊(r,t) in the two regions adopt opposite signs with respect to m₊(r,t). Consequently, B_d,+(r,t) tilts m(r,t) in the inner region toward the +z-axis while rotating m(r,t) in the outer region toward -z. This enhances the difference in B_d,_z(r,t) as well as the corresponding precession frequencies of m(r,t) between the two regions. As a result, the phase of B_d,+(r,t) with respect to m₊(r,t) increases in each region, forcing mjj,t) even farther apart by positive feedback. The linear instability of the original magnetization configuration under the DDF produces exponentially increasing contrast in time, as measured through Δm_z(t) (Δm_z(t)≡<m_z(r_out,t)>-<m_z(ri_n,t)>, averaged along the z-axis).

By comparison, in the radiation damping regime (FIG. 2B), following a large flip- angle pulse (Θ= 179°), the magnetization is nearly tipped into the unstable inverted state. The resulting B^t) triggered by the small net transverse magnetization <m₊> is initially not on resonance with the magnetization in either region (see, Eq. 4), since m₊(r,t) in the two regions precess at different frequencies. However, B_r(t) evolves to be more closely on-resonance with <w₊> in the region with greater total magnetization, which has a greater weighted contribution to <m₊>. As a result, m(r,t) in this region nutates toward the +z-axis more quickly, creating still more transverse magnetization. B^t) in turn excites the spins in this same region even more effectively while m(r,t) in the other region remains comparatively off-resonance with respect to B/t). The increasing selectivity of B_r(t) is reflected in FIG. 2B, which shows that B_r(t) is highly selective for resonance frequency differences δω as small as 5 Hz, with the resulting contrast growing as a power law in time. Such avalanching amplification of MRI contrast by radiation damping is demonstrated experimentally on samples of human brain tissue for lesion characterization (FIGs. 3-9). For very small differences in resonance frequency, the dynamics under the joint feedback fields of radiation damping and the DDF may provide even better contrast enhancement than that generated by either feedback field alone. Recent studies have revealed that radiation damping and the DDF combine to generate dynamical instability (as described further in J. Jeener, J. Chem. Phys. 116, 8439 (2002), the disclosures of which are incorporated herein by reference) leading to chaotic spin dynamics (as described further in Y.-Y. Lin, et. al., Science 290, 118 (2000), the disclosure of which is incorporated herein by reference) in high-field MR experiments. The spin dynamics responsible for contrast enhancement under the joint feedback fields may be understood as follows. After a [Θ_X-(GT)_Z] preparation sequence, m(r,t) evolves primarily under B^r,t) at short times, since the dephasing gradient spoils <m₊>, rendering B_r(t) weak. The nonuniform distribution of resonance frequencies causes m₊(r,t) in both regions to deviate from <m₊>. As a result, B_d,+(r,t) follows the precession of m₊(r,t), and its phase with respect to m₊(r,t) assumes different signs in each region (Eq. 2). B_{d +}(r,t) thus tilts m_z in opposite directions in the two regions, in close analogy with the mechanism shown in FIG. 2A. At the same time, a small residual <m₊> surviving the crusher gradient activates B_r(t). B_r(t) acts on m(r,t) to produce a modulation in m/r,rj, which in turn triggers the DDF to distort the magnetization helix and refocus more <m₊> (S. Y. Huang et. al., J. Chem. Phys. 121, 6105 (2004), the disclosure of which is incorporated herein by reference). The effect of the joint reaction fields is reinforced on the region with greater total magnetization (FIG. 2C, bottom left), while the other frequency component remains largely off -resonance with respect to the joint reaction fields. The feedback fields intensify until <m₊> reaches a maximum, accompanied by the steep rise in Δm_z (FIG. 2C, top left). Contrast enhancement by the joint feedback fields is confirmed through experimental results for a simple water/ethanol phantom (FIG. 2C, right).

The choice of which avalanching amplification mechanism of the current invention (as shown and discussed in relation to FIG. 2) to use experimentally depends on the contrast origin of interest (e.g., proton density or resonance frequency), reaction field strengths, and physical constraints (e.g., relaxation times). The short T₂ relaxation times in biological systems (-0.1 s) favor the mechanism depicted in Fig. 2B: avalanching amplification of small variations in resonance frequency by radiation damping. While the sensitivity and quality factor of the RF receiver coil in most MR spectrometers/scanners is not high enough to induce a strong radiation damping field for in vivo imaging, utilizing an external electronic device can significantly enhance the radiation damping feedback field and provide an opportunity to design new imaging pulse sequences in which the feedback interaction is controllable. The radiation damping feedback field may be amplified for imaging at the lower field strengths used in conventional MRI through electronic feedback to the induced circuit using modified probes, as outlined in FIG. 3.

As a demonstration, Applicants applied avalanching amplification of MRI contrast under radiation damping to improve tissue differentiation in two types of brain lesions (FIGs. 4-10) and in vivo fish (FIG. 12) and frog embryos (FIG. 13) imaged at 14.1 T. In vivo mice (FIG. 14) were also imaged at 7 T.

EXAMPLES:

The examples set forth in the following sections contain MR images that were obtained using brain tissues excised from patients undergoing intracranial surgeries as well as in vivo images acquired on common guppy fish and live frog embryos. In vivo feedback- enhanced imaging was also demonstrated on mice using a modified electronic circuit. The tissue experiments were performed on brain tissues that were not needed for immediate pathology examinations. Risks and benefits of the experimental procedure were explained to the adult patients and the guardians of the pediatric patients by the attending neurosurgeon and the investigators. All patients (or their guardians) in this study gave written informed consent, and the research protocols were approved by the UCLA Institutional Review Board (IRB). The ages of the patients were 51 (white male) and 6 years (Hispanic male).

The pediatric patient was evaluated with a detailed clinical history, neurological examinations, electroencephalography (EEG), and neuroimaging with high-resolution MRI and ¹⁸fluoro-2-deoxyglucose positron emission tomography (PET). The epileptogenic region for surgical resection was anatomically defined based on convergent EEG and neuroimaging abnormalities (for procedure see G. W. Mathern et al., Epilepsia 40, 1740 (1999), the disclosure of which is enclosed herein by reference). At surgery, electrocorticography further defined the brain regions to be removed (for a further discussion see G. W. Mathern et al., Epilepsia 41, S 162 (2000); and C. Cepeda et al., /. Neurosci. Res. 72, 472 (2003), the disclosures of which are incorporated herein by reference). The pediatric patient underwent hemispherectomy for cortical dysplasia involving a large area of the left hemisphere. The tissue sample was obtained from the left posterior parietal-occipital lobe. The adult patient underwent surgery for resection of glioblastoma multiforme (GBM) located in the left temporal lobe. The diagnosis and classification of the tumor was confirmed through presurgical biopsy and postsurgical pathological examination. The patient did not receive any radiation therapy prior to resection of the tumor. Pathological examination of the tumor tissue confirmed high-grade glioblastoma with focal extension into the subarachnoid space, areas of hypervascularization, and areas of geographic and pseudopalisading necrosis hallmarked by condensation of tumor cells.

In each case, a 5 mm-diameter block of the larger resection was excised by the neurosurgeon and placed in a 5-mm MR sample tube filled with 0.9% sodium chloride solution. Samples were maintained at 5°C to preserve tissue integrity for up to 24 hours. Following completion of MR studies, tissue samples were immediately fixed in 4% paraformaldehyde for 5 days, cryoprotected for two nights in increasing sucrose concentrations (20-30%) diluted in phosphate-buffered saline, frozen, and stored at -80⁰C. Cryostat sections (30 μm) were rinsed in Tris-saline, mounted on gelatin-coated slides, and air-dried. The slides were processed the next day as follows: 60 min in chloroform to remove lipids, 5 min each in 100%, 95%, and 75% alcohol and water for rehydration, and 5 min in 0.1% cresyl violet stain with acetic acid buffer. After sufficient coloring, the slides were dehydrated through immersion in 75%, 95%, and 100% alcohol for 5 min each, dipped in xylene for 10 min, and coverslipped.

In vivo feedback-enhanced images were acquired of common guppy fish, Poecilia reticulata (P. reticulata), obtained from commercial suppliers in California. The fish were maintained in filtered water at 290 K. The temperature of the water was kept at 290 K to slow down movement of the fish while maintaining survival, as monitored experimentally. Subjects were then placed in 5 mm MR sample tubes with filtered water for MR measurements. Guppy fish that were slightly smaller in diameter than the tubes were chosen for the experiments to ensure a secure fit, such that neither immobilization nor anesthesia was needed. The sagittal profile of the fish in FIG. 11 shows how the fish was positioned tightly within the tube while allowing room for the fins to jut out to the side (left fin not seen in this slice).

Xenopus laevis (X. laevis) frog embryos were obtained from Nasco (Fort Atkinson, Wisconsin) 12 hours after in vitro fertilization and were maintained in filtered water at 289 K. Subjects were then transferred to 5 mm MR sample tubes with filtered water for imaging experiments. Twelve embryos were placed in each tube, which allowed for tight packing due to the close association of adjacent embryos suspended within their respective jelly-filled sacs. The temperature was kept at 289 K (well within the temperature range required for proper development) to reduce the rate of growth of the embryos over the time needed to acquire the images.

In vivo and in vitro images were acquired using a Bruker Avance 600 MHz (Bruker BioSpin Corp., Billerica, MA) microimaging system equipped with a narrow-bore (54 mm) 14.1-T magnet and Micro5 gradient system (maximal gradient strength of 192 G/cm in three orthogonal directions). A 5-mm saddle coil optimized for ¹H sensitivity was used for RF transmission and reception. Detailed description of imaging sequence parameters and data processing are provided in the descriptions of the images provided below.

To evaluate quantitatively the performance of feedback-based contrast enhancement, contrast-to-noise ratios (CNRs) were calculated by taking the difference of the mean signal intensities in designated regions of interest (16 pixels square each) for the radiation damping- enhanced MR magnitude images in the figures and dividing this difference by the noise, sampled in regions of interest comprising no signal (for further information on this technique see S. D. Wolff, R. S. Balaban, Radiology 202, 25 (1997), the disclosure of which is incorporated herein by reference). CNRs were also calculated for the corresponding T ₂- weighted MR images, which had the best contrast among the conventional images.

For the in vivo experiments on mice, male ICR mice weighing 36.68 ± 7.11 g (mean standard deviation (SD)) were provided by the Division of Pulmonary and Critical Care Medicine at the Tri-Service General Hospital in Taipei, Taiwan. The mice were anesthetized with gaseous anesthesia using an initial dose of 2.0% isofluorane in air and a small animal gating system (SA instruments Inc., NY, USA). For maintenance, the isofluorane was set under 1.0% and gated with a respiration trigger sensor. The respiration rate of the mice was controlled under 60 breaths per minutes. Warm air at 28 ± 2°C was transported to the mice to avoid loss in temperature and was regulated by a rectal temperature probe (SA Instruments Inc., NY, USA). All in vivo MR images of mice were acquired using a Varian INOVA 7 T NMR spectrometer (Varian, CA, USA) with microimaging capability. The images were obtained using a microimaging probe head (Resonance Research Inc., Billerica, MA, USA), which comprises a quadrature birdcage imaging RF coil (30 mm LD.) and a self-shielded gradient system with a maximum strength of 100 gauss/cm in each of the x-, y- and z-directions. A block diagram of the active electronic feedback circuit is shown in FIG 3. Active feedback was switched on during the evolution time τ, sandwiched between the first 180° hard pulse and first slice-selective soft pulse. Sagittal active feedback-enhanced images were acquired using the following imaging parameters: TE = 9.5 ms, TR = 5 sec, FOV = 60 mm x 30 mm, acquisition matrix = 128 x 64, slice thickness = 1 mm. The duration of the nonselective 180° pulse was 51 μs.

Example 1 : Cortical Dysplasia Images

The first case shown in FIGs. 4 and 5 involved focal cortical dysplasia, which is linked to medically intractable epilepsy (D. C. Taylor et al., /. Neurol. Neurosurg. Psychiatry 34, 369 (1971), the disclosure of which is incorporated herein by reference). Cortical dysplasia is characterized histopathologically by cortical laminar disorganization and blurring of the gray and white matter junction.

FIGs. 4A to 4D show the following images: (FIG. 4A) Radiation damping-enhanced (RD) and (FIG. 4B) T₂-weighted (T₂) MR images at 14.1 T of brain tissue excised from the left posterior parietal-occipital lobe of a pediatric patient with cortical dysplasia, compared with (FIG. 4C) histopathology (Hist.) and (FIG. 4D) gross anatomy (Anat.). Feedback- enhanced images show amplified contrast between gray matter (asterisk) and white matter (arrowhead) in mildly dysplastic tissue, with corresponding contrast-to-noise ratios (CNRs) of 60.5 in (FIG. 4A) and 4.2 in (FIG. 4B). Field distortion due to iron in hemoglobin creates imaging artifacts surrounding the blood vessels in (FTG. 4A and FIG. 4B) (arrow). Pulse sequence at top was used in (FIG. 4A) to enhance MRI contrast through the radiation damping feedback field (Gr=5G-ms/cm). In (FIG. 4A and FIG. 4B), the sample was imaged by a gradient-echo sequence (vertical 1-mm thick slice, 512 x 128 voxels zero-filled to 512 x 256 voxels, field of view 1.6 cm, echo time (TE) of 4.2 ms in (FIG. 4A) and 20 ms in (FIG. 4B)). MR images of all tissue samples were brightened through nonlinear scaling of the color map intensity without changing the actual signal intensity values.

Scale bars and additional MR images for comparison are shown in FIGs. 4A to 4D. Comparison of (FIG. 5A) radiation damping-enhanced (RD) and conventional (FIG. 5B) T₂-, (FIG. 5C) Tj-, and (FIG. 5D) proton density (PD) MR images of brain tissue excised from the left posterior parietal-occipital lobe of a pediatric patient with cortical dysplasia, corresponding to those shown in FIGs. 3. All four sequences identified the area of blood (arrow, — >) in the brain tissue. However, only the radiation damping-enhanced image differentiated gray (asterisk, *) from white matter (arrowhead, Δ). In (FIG. 5A), the radiation damping-enhanced image was acquired with the preparation sequence shown in FIG. 4 (#=175°, Gr=5G-ms/cm, £=75 ms, TE=4.2 ms, TR=5 s). In (FIG. 5B), the T₂-weighted image was acquired with TE=20 ms and TR=5 s. In (FIG. 5C), the Ty-weighted image was acquired following the sequence shown in Fig. 3 (initial flip angle #=180°, similar to an inversion recovery experiment) with a weak gradient of strength G_z=0.048 G/cm applied during the τ evolution time to spoil the transverse magnetization in order to eliminate radiation damping (£=1.5 s, TE=4.2 ms, TR=5 s). In (FIG. 5D), the proton density image was acquired with TE=4.2 ms and TR=5 s. In (FIGs. 4A-D), the sample was imaged by a gradient-echo sequence (vertical 1-mm thick slice, 512 x 128 voxels zero-filled to 512 x 256 voxels, field of view 1.6 cm). All MR images were brightened through nonlinear scaling of the color map intensity without changing the actual signal intensity values. Scale bars are given in arbitrary units.

Reviewing the above-discussed figures, FIG. 4 (detailed in FIG. 5) compares radiation damping-enhanced and conventional ^-weighted MR images of mildly dysplastic in vitro unfixed brain tissue excised from the left posterior parietal-occipital lobe of a patient with cortical dysplasia. The field distortion due to the presence of iron in hemoglobin creates imaging artifacts surrounding the blood vessels (FIG. 4, arrow). While the ^-weighted image (FIG. 4B/5B) provides better contrast than the proton density (FIG. 5D) or Ty-weighted images (FIG. 5C), the conventional images fail to differentiate the gray and white matter. In comparison, the radiation damping-enhanced image shows a clear change in contrast at the junction between the gray and white matter (Fig. 4A), with an increase in contrast-to-noise ratio (CNR) of about 15 times compared to the ^-weighted image. The radiation damping field following the initial 175° pulse selectively excites the magnetization in different regions based on resonance frequency differences reflecting inherent variations in magnetic susceptibility, which arise from the different levels of deoxyhemoglobin and cerebral blood volume in gray and white matter. (Bartha, R., Michaeli, S., Merkle, H., Adriany, G., Andersen, P., Chen, W., Ugurbil, K. & Garwood, M. (2002) Magn. Reson. Med. 47, 742- 750, the disclosure of which is incorporated herein by reference). Such gray-white matter differentiation using this novel MRI approach could be used to identify subtle malformations in cortical development.

Example 2: GBM Images

Glioblastoma multiforme (GBM) is the most common malignant primary brain tumor. Two important histopathological features of GBM are necrosis and microvascular proliferation (as discussed further in P. Kleihues et ah, in Pathology and Genetics of Tumours of the Nervous System, P. Kleihues, W. K. Cavenee, Ed. (IARC Press, Lyon, 2000), pp. 29- 39, the disclosure of which is incorporated herein by reference). These features are reflected in microscopic MR images of three separate brain tissue sections taken from the left temporal lobe of a patient with GBM, shown in FIGs. 6, 7 (detailed in FIG. 8), and 9 (detailed in Fig. 10).

FIGs. 6A to 6E provide MR images for comparison of (FIG. 6A) histopathology with (FIG. 6B) radiation damping-enhanced and (FIG. 6C) conventional Ti-, (FIG. 6D) Tj-, and (FIG. 6E) proton density MR images of brain tissue excised from the left temporal lobe of an adult patient with GBM. Comparison with histopathology shows that areas of necrosis (arrowhead, Δ), tumor tissue (asterisk, *), and tumor tissue interspersed with necrosis (arrow, T) are highlighted in the MR images, confirming the agreement between features seen in feedback-enhanced and conventional images. In (FIG. 6A), the tissue section was magnified by 2Ox. In (FIG. 6B), the radiation damping-enhanced image was acquired following the preparation sequence shown in FIG. 4

TR=5 s). In FIG. 6C, the T₂-weighted image was acquired with TE=20 ms and TR=3 s. In FIG. 6D, the Ty-weighted image was acquired following preparation by the sequence shown in FIG. 4 (initial flip angle #=180°, similar to an inversion recovery experiment) with a weak gradient of strength G_z=O.048 G/cm applied during the τ evolution time to spoil the transverse magnetization in order to eliminate radiation damping (£=0.8 s, TE=4.2 ms, TR=3 s). In FIG. 6E, the proton density image was acquired with TE=4.2 ms and TR=3 s. In FIG. 6B-E, images of the upper and lower halves of the tissue sample were taken separately using gradient-echo imaging (vertical 1-mm thick slices, 512 x 128 voxels zero-filled to 512 x 256 voxels, field of view 0.8 cm each, total length 1.45 cm), then merged at their interface (indicated by the dashed line). All MR images were brightened through nonlinear scaling of the color map intensity without changing the actual signal intensity values. Scale bars are given in arbitrary units

FIGs. 7A to 7C provide (FTG. 7A) histopathology, (FIG. 7B) radiation damping- enhanced, and (FIG. 7C) ^-weighted MR images of brain tissue excised from the left temporal lobe of the same adult patient with GBM. The radiation damping-enhanced image (FIG. 7B) differentiates tumor (asterisk) versus necrotic brain regions (arrowhead) (CNR=55), while the T₂-weighted image (FIG. 7C) does not (CNR=2.4). Blood vessel cross- sections are seen adjacent to the tumor (arrow). In FIGs. 7A and 7B, images of the upper and lower halves of the tissue sample were taken separately (field of view 0.8 cm each, total length 1.4 cm), then merged at their interface (indicated by the dashed line). Pulse sequence, parameters, and MR image processing are the same as in FIG. 4.

Scale bars and additional MR images for comparison with those shown in FIG. 7 are shown in FTG. 8. Specifically, FIGs. 8A to 8E provide a comparison of (FTG. 8A) histopathology with (FIG. 8B) radiation damping-enhanced and (FIG. 8C) conventional T₂-, (FTG. 8D) Ti-, and (FIG. 8E) proton density MR images, corresponding to the images shown in Fig. 7, of another brain tissue section excised from the left temporal lobe of the same adult GBM patient as in FIGs. 6, 7, and 10. The radiation damping-enhanced image (FTG. 8B) differentiates between tumor (asterisk) versus necrotic brain regions (arrowhead), while the other MR images do not. Blood vessel cross-sections are seen adjacent to the tumor (arrow, <— ). In (FIG. 8A), the tissue section was magnified by 2Ox. In (FIG. 8B), the radiation damping-enhanced image, prepared with sequence shown in FIG. 4 (#=175°, G7=5G-ms/cm,"£=75 ms, TE=4.2 ms, TR=5 s). In (FIG. 8C), the T₂-weighted image was acquired with TE=20 ms and TR=3 s. In (FIG. 8D), the 7_/-weighted image was acquired following the preparation sequence shown in FIG. 4 (initial flip angle #=180°, similar to an inversion recovery experiment) with a weak gradient of strength G_z=0.048 G/cm applied during the τ evolution time to spoil the transverse magnetization in order to eliminate radiation damping ( 7=0.8 s, TE=4.2 ms, TR=3 s). In (FIG. 8E), the proton density image was acquired with TE=4.2 ms and TR=3 s). In (FIGs. 8B-E), images of the upper and lower halves of the tissue sample were taken separately using gradient-echo imaging (vertical 1-mm thick slices, 512 x 128 voxels zero-filled to 512 x 256 voxels, field of view 0.8 cm each), then merged at their interface (indicated by the dashed line). All MR images were brightened through nonlinear scaling of the color map intensity without changing the actual signal intensity values. Scale bars are given in arbitrary units

As shown in the above-discussed figures, areas of necrosis caused by vascular thrombosis appear hyperintense in proton density images due to the breakdown of proteinaceous tumor tissue (as discussed more completely in R. A. Zimmerman, L. T. Bilaniuk, in Neuroimaging: Clinical and Physical Principles, R. A. Zimmerman, W. A. Gibby, R. F. Carmody, Ed. (Springer, New York, 2000), pp. 986-991, the disclosure of which is incorporated herein by reference). The rupture of poorly formed tumor vessels enhances signal intensity in Ti- and T ₂- weigh ted images by releasing paramagnetic methemoglobin into the extracellular necrotic zone. These features are observed in proton density, Ti-, and T₂- weighted images of the tumor sample in FIG. 6, as well as in the corresponding radiation damping-enhanced image and histopathology. On the other hand, the conventional images of another tumor sample, shown in FIG. 7 (detailed in FIG. 8), do not reflect the difference between necrotic and tumor tissues seen in pathology (FIG. 7A). In comparison, the radiation damping-enhanced image (FIG. 7B) distinguishes between the two regions, enhancing the CNR by 24 times over the ^-weighted image. The radiation damping-enhanced image is sensitive to resonance frequency differences arising from bulk magnetic susceptibility (χ) variations in hypointense, actively dividing tumor cells (varying amounts of oxyhemoglobin and deoxyhemoglobin, Δχ_^-0.0lAl ppm and 0.265 ppm, respectively, relative to bulk water) and hyperintense necrotic tissue (extracellular methemoglobin, 43^=0.301 ppm) (as disclosed in M. Zborowski et al, Biophys. J. 84, 2638 (2003), the disclosure of which is incorporated herein by reference). Necrosis is considered to be an anaplastic feature of astrocytoma and is associated with a poorer prognosis. By providing a better distinction between necrotic and tumor tissues, feedback-enhanced MR contrast may be used in preclinical studies to develop criteria for characterizing the appearance of malignant brain tumors without resorting to surgical biopsy.

Similarly, FIGs. 9A to 9D provide, (FIG. 9A) radiation damping-enhanced MR image, (FIG. 9B) histopathology, (FIG. 9C) r₂-weighted MR image, and (FIG. 9D) gross anatomy of another brain section taken from the same GBM patient as in FIG. 7. Regions of enhanced signal intensity in (FIG. 9A) correspond to tumor cells surrounding the microvasculature (red circle) compared to normal tissue (yellow circle) (CNR=42.6 in (FIG. 8A) versus CNR=2.2 in (FIG. 9C)). Comparison with histopathology (FIG. 9B) reveals that regions of enhanced signal intensity correspond to tumor cells surrounding the microvasculature (red circle) compared to normal tissue (yellow circle). Blood vessels are seen running longitudinally through this slice (arrow, — >). Pulse sequence, parameters, and MR image processing are the same as in FIG. 4, except field of view is 1.2 cm (only 1 cm is shown here).

Scale bars and additional MR images for comparison are shown in FIG. 10. Specifically, FIGs. 1OA to 1OD provide a comparison of (FIG. 10A) radiation damping- enhanced and (FIG. 10B) conventional T₂-, (FlG. 10C) T₁-, and (FTG. 10D) proton density MR images, corresponding to those shown in FIG. 9, of the third brain tissue section excised from the left temporal lobe of the same adult GBM patient as in FIGs. 6 and 8. In (FIG. 10A), the radiation damping-enhanced image was acquired following the preparation sequence shown in FIG. 4 (#=175°, Gr=5G-ms/cm,£=75 ms, TE=4.2 ms, TR=5 s). In (FTG. 10B), the ^-weighted image was acquired with TE=20 ms and TR=3 s. In (FTG. 10C), the T_/-weighted image was acquired following the preparation sequence shown in FTG. 4 (initial flip angle #=180°, similar to an inversion recovery experiment) with a weak gradient of strength G_z=0.048 G/cm applied during the revolution time to eliminate radiation damping (τ=0.8 s, TE=4.2 ms, TR=3 s). In (FIG. 10D), the proton density image was acquired with TE=4.2 ms and TR=3 s. In (FIG. lOB-D), images were acquired by gradient-echo imaging (vertical 1- mm thick slices, 512 x 128 voxels zero-filled to 512 x 256 voxels, field of view 1.2 cm, with only 1 cm shown here). All MR images were brightened through nonlinear scaling of the color map intensity without changing the actual signal intensity values. Scale bars are given in arbitrary units.

As shown in the above figures, the radiation damping feedback field can also be used to distinguish tumor growth from surrounding healthy tissue. Tumor cells surrounding necrosis are intimately involved in the proliferation of microscopic blood vessels. Cross- sections of hyperplastic vasculature appear adjacent to the tumor in FIG. 7 (detailed in FIG. 8), and blood vessels are seen running longitudinally in another section of the same tumor shown in FIG. 8 (detailed in FIG. 10). Paramagnetic deoxyhemoglobin in residual blood manifests in ^-weighted images as signal loss (FIG. 9C) and produces hyperintensity due to variations in resonance frequency in the corresponding radiation damping-enhanced image (FIG. 9A). Furthermore, tumor cells surrounding the microvasculature are highlighted in the radiation damping-enhanced image, corresponding to an increase in CNR of 20 times over the ^-weighted image. The hyperintense regions in FIG. 9A correspond to differences in bulk susceptibility originating from variations in blood oxygenation level and increased water content in the compact extracellular space of the tumor. These clusters of malignant cells are not obvious in the proton density image and may not be sufficiently vascular to enhance on Tj- or ^-weighted images.

Example 3: Epileptogenic Lesions

In addition to the above images, a systematic study of 25 tissue samples from epileptogenic lesions in five patients was examined. The images thus taken showed improvement in gray/white matter contrast by six times on average under radiation damping over the best of optimized conventional Ti -weighted, T₂-weighted, and proton density images. These results are presented in FIG. 11. Specifically, in FIG. 11, contrast-to-noise ratios for radiation damping-enhanced and conventional Tj -weighted, T₂- weighted, and proton density images are reported. Average CNRs are reported in the figure legend as mean ± standard deviation for 25 sections of brain tissue in total obtained from five patients with epileptogenic lesions. These results demonstrate that gray-white matter differentiation using this novel MRI approach can be applied to identify subtle malformations in cortical development.

Example 4: In vivo Guppy Fish.

Contrast enhancement under radiation damping and the joint feedback fields highlighted anatomical features of a live guppy fish not discernable in the corresponding conventional T₂ ^*-weighted, 7Vweighted, and proton density images. FIG. 12 compares images of a slice within the head of the fish obtained by different imaging methods. In the simulated and experimental radiation damping-enhanced images (FIGs. 12A and C), regions corresponding to the eyes appeared bright against the darker facial tissue, while in the simulated and experimental joint feedback field-enhanced images shown in FIGs. 12B and D, the eyes appeared darker than the surrounding facial tissue. In comparison, conventional T₂ - weighted, 7\-weighted, and proton density images did not show noticeable contrast between the eyes and the adjacent tissue.

Example 5: In vivo Frog Embryos FIG. 13 shows experimental MR images comparing contrast enhancement under radiation damping alone with contrast enhancement under radiation damping and additional RF pulses on a developing X. laevis embryo. The detailed internal structure of the embryo appeared to be more complex than that of the guppy fish; thus, simulations on the embryo sample were not carried out to reproduce the observed contrast. Nevertheless, the spin dynamics and demonstration of contrast enhancement due to radiation damping could be roughly considered in terms of two broad regions of the embryo: the head and the tail. Following an initial 175° pulse, radiation damping-enhanced images were acquired for various free evolution times τ. At τ = 50 ms (FIG. 13A), the dynamics under radiation damping approached a steady state, and the contrast between the tail and head regions of the embryo was improved compared with that seen in conventional T₂ ^*-weighted, T_\ -weighted, and proton density images. The CNR between the eyes (marked "1" in the figure) and gills (marked "3") was enhanced by about three times under radiation damping (FIG. 13C, CNR = 1.84) compared with the T₂ ^*-weighted image (FTG. 13D, CNR = 0.66), while the CNR between the ear vesicles (marked "2") and the gills (marked "3") was enhanced by 2.6 times under radiation damping (FIG. 13C, CNR=2.63) compared with the T₂ ^*-weighted image (FIG. 13D, CNR=LOl). Since the magnetization near the head may have had a smaller resonance offset with respect to the surrounding water, the magnetization in this region rotated past the transverse plane more quickly under radiation damping and was brighter than the magnetization in the tail region, which remained closer to the transverse plane.

Example 6: In vivo Mouse Imaging

An active RF feedback circuit has been designed to amplify and control the radiation damping feedback field after a 175° excitation pulse and imaged the brains of mice in vivo [FIG. 14]. Active feedback-enhanced images were able to highlight tissue boundaries that were not distinct in the corresponding conventional proton density, Tl -weighted, and T2*- weighted images [FIGs. B-D]. Analysis of the underlying dynamics suggests that the boundary enhancement results from the strong RD feedback field acting on tissue regions with different magnetic susceptibility and superimposed background field inhomogeneity. The RD field acts as a highly selective self-induced RF field to differentiate tissues with distinct resonance frequencies. Inhomogeneity across different tissue regions can serve as an endogenous encoding gradient to reinforce the resonance offset between such tissues. The RD field then acts like a soft slice selective pulse to highlight the interface between the tissues. Differential excitation under the feedback field thereby distinguishes the tissues and enhances contrast at the tissue boundaries. The development of an active feedback circuit to amplify the RD field thus enables improved differentiation of neighboring tissues at low fields using conventional probes/receiver coils.

The above figures demonstrate avalanching amplification of MRI contrast due to small differences in spin density or resonance frequency under the feedback interactions of the distant dipolar field and/or radiation damping in phantoms and in vitro human brain tissue. Observations show up to 20 times improved contrast in epileptogenic lesions, e.g., cortical dysplasia (FIGs. 4 and 5), and malignant brain tumors, e.g., glioblastoma multiforme (FIGs. 6 to 10), tissues with minimal contrast differences in routine MRI. In evaluating how robust feedback-based contrast enhancement is to noise, it should be noted that avalanching amplification by the individual or joint feedback fields causes image contrast to grow rapidly before reaching a steady, significant value. During the nonlinear amplification process, larger differences in MR parameters that reflect the underlying anatomy or structures are highlighted first. In the absence of relaxation, small fluctuations due to noise or non-uniform experimental perturbations are eventually enhanced at long times. Knowing the stability of magnetization configurations under the reaction fields can guide the design of RF pulse sequences that further amplify desired contrast while suppressing the growth of noise. It should be understood that such advancements are further contemplated by the current invention. The simple preparation sequences shown here belie the complexity of the underlying dynamics and introduce a novel approach to designing MR pulse sequences. In this new approach, evolution under the reaction fields allows the spins themselves to play an active role in determining and differentiating their subsequent evolution, thereby improving the distinction between regions with different MR properties. For example, contrast in the feedback-enhanced images (FIGs. 4 to 10) exceeds that seen in conventional chemical-shift selective images (results to be reported elsewhere), which fail to distinguish limited differences in precession frequency in the presence of the large background field inhomogeneity present in heterogeneous samples. Although the above discussion has focused on the imaging of only a few types of physiological conditions, it should be understood that the feedback-based contrast enhancement system and method of the current invention is sensitive to small differences in endogenous MR parameters that correlate with versatile physiological origins. For example, gray and white matter in dysplastic tissue are differentiated through amplified contrast sensitive to the increased level of compartmentalized ferritin and blood volume in gray matter compared with white matter, while limited variations in the concentrations of oxyhemoglobin, deoxyhemoglobin, and methemoglobin in healthy, tumor, and necrotic tissues are highlighted based on differences in magnetic susceptibility. Other potential applications include: enhancing functional MRI (fMRI) contrast based on the blood oxygenation level-dependent (BOLD) effect; distinguishing the penumbra zone of stroke (i.e., reversibly injured tissue) from irreversible infarction through amplified pH-dependent contrast (see, K. M. Ward, R. S. Balaban, Magn. Reson. Med. 44, 799 (2000); and J. Y. Zhou, J. F. Payen, D. A. Wilson, R. J. Traystman, P. C. M. van Zijl, Nat. Med. 9, 1085 (2003) the disclosures of which are incorporated herein by reference); and improving tumor characterization by sensitive detection of amide-proton exchange rate variations between normal and cancerous tissue (see, S. D. Wolff, R. S. Balaban, J. Magn. Reson. 86, 164 (1990); and J. Y. Zhou, B. LaI, D. A. Wilson, J. Laterra, P. C. M. van Zijl, Magn. Reson. Med. 50, 1120 (2003), the disclosures of which are incorporated herein by reference). The observed illumination of paramagnetic hemoglobin in the feedback-enhanced images suggests that this amplification scheme may also be integrated with targeted ferromagnetic nanoparticles in molecular imaging to achieve better disease specificity and lower detection limits of molecular events (see, J. M. Tyszka, S. E. Fraser, R. E. Jacobs, Curr. Opin. Biotechnol. 16, 93 (2005), the disclosure of which is incorporated herein by reference).

Discussion of System Although the above discussion has focused on a method of enhancing NMR and MR detection using a feedback methodology, the invention is also directed to systems of implementing this methodology. Such implementation can include specially constructed MR devices, or accessory attachments for retro-fitting conventional MR devices to utilize the inventive method.

The feedback interactions discussed here become more pronounced under conditions developed for high-sensitivity MR imaging and microscopy, i.e., high fields, sensitive probes, and/or highly polarized samples. Such feedback fields are thus readily adapted for contrast enhancement in in vitro and in vivo preclinical studies by MR microscopy. The use of feedback fields may also be generalized to lower fields or clinical scanners through careful consideration of the experimental system, pulse sequence design, and imaging hardware.

For example, the DDF has been applied in MRI to generate imaging contrast via intermolecular multiple quantum coherences down to field strengths of 1.5 T in clinical scanners. (Such systems are discussed more fully in W. S. Warren et al., Science 281, 247 (1998); and J. H. Zhong, Z. Chen, E. Kwok, Magn. Reson. Med. 43, 335 (2000), the disclosures of which are incorporated herein by reference.) Lower field strengths in general prolong T₂ relaxation times, allowing the DDF to act over longer evolution periods. The use of dipolar instabilities to amplify spin precession signals in systems with high spin polarization suggests that such instabilities may also be applied to enhance contrast in low- field MRI with hyperpolarized spins. (See, e.g., M. P. Ledbetter, I. M. Savukov, M. V. Romalis, Phys. Rev. Lett. 94, 060801 (2005); and C. H. Tseng et al., Phys. Rev. Lett. 81, 3785 (1998), the disclosures of which are incorporated herein by reference.) Regarding radiation damping, this feedback field has been observed in in vitro and in vivo imaging experiments as low as 4.7 T. (J. Zhou, S. Mori, P. C. M. van Zijl, Magn. Reson. Med. 40, 712 (1998), the disclosure of which is incorporated herein by reference.) Hardware modifications can also bolster the radiation damping effect at low fields. For example, high-β and cryogenic probes with short τ_r values are used in MR microscopy and clinical imaging at both high and low fields. (For a further discussion, see A. C. Wright, H. K. Song, F. W. Wehrli, Magn. Reson. Med. 43, 163 (2000); and Q. Y. Ma et al, Acad. Radiol. 10, 978 (2003), the disclosures of which are incorporated herein by reference.) Radiation damping can also be heightened or suppressed through active electronic feedback to the induced current in modified circuits, which are already commercially available. (See, e.g., P. Broekaert, J. Jeener, J. Magn. Reson., Ser. A 113, 60 (1995); and A. Louis-Joseph, D. Abergel, J.-Y. Lallemand, /. Biomol. NMR 5, 212 (1995), the disclosures of which are incorporated herein by reference.) Contrast enhancement under radiation damping could be envisioned in clinical MR scanners by applying adiabatic pulses to invert the magnetization uniformly and partially counter B₁ inhomogeneity effects (see, Garwood M, DelaBarre L. The return of the frequency sweep: Designing adiabatic pulses for contemporary NMR. J Magn Reson 2001;153:155-177, the disclosure of which is incorporated herein by reference), then enhancing the radiation damping field through a radiation damping control unit. Furthermore, as radiation damping is simply an RF field generated by the receiver coil through the spins, its effect may be mimicked or even improved upon through the design of complex continuous-wave pulse sequences in cases when the radiation damping field is weak or absent.

Summary The approach of the current invention yields robust image contrast sensitive to small differences in endogenous MR parameters that is not seen in conventional MR images. For example, the spin dynamics under radiation damping amplify contrast due to slight variations in resonance frequency, which are difficult to distinguish by existing methods. Frequency selective excitation or saturation methods are highly susceptible to poor field homogeneity and exhibit unexpectedly complex dynamics in the presence of radiation damping. Furthermore, susceptibility-weighted imaging methods that measure phase differences between different frequency components are not sufficiently sensitive to very small resonance frequency differences. On the other hand, the exquisite selectivity of the dynamics under radiation damping makes this method more robust to static field inhomogeneity than conventional selective pulses, and heightens contrast between tissues with small differences in resonance frequency reflecting the surrounding chemical or magnetic environment, all within a single image. Furthermore, the simplicity of the preparation sequences used to enhance contrast under the feedback fields facilitates incorporation of such methods into existing pulse sequences with little modification. Although useful in a wide variety of MR uses, the contrast enhancement provided by nonlinear feedback of the current invention is especially important in biomedical applications of MRI, where limited contrast often makes it difficult to detect the physiological changes leading to a pathologic state and to delineate the exact extent of lesions such as small tumors. Such avalanching amplification of MRI contrast has been shown herein to improve the characterization of epileptogenic lesions and malignant brain tumors, which are notoriously difficult to visualize accurately even with current, state-of-the-art MRI. Feedback-based contrast enhancement enables more precise delineation of indistinct features in a variety of biomedical, tissue, and materials MR imaging applications, making it appealing to a wide range of physical, chemical, and biomedical scientists, all of which are contemplated by the current disclosure.

Accordingly, while preferred embodiments of the foregoing invention have been set forth for purposes of illustration, the foregoing description should not be deemed a limitation of the invention herein. Accordingly, various modifications, adaptations and alternatives may occur to one skilled in the art without departing from the spirit and scope of the present invention.

Claims

WHAT IS CLAIMED IS:

1. A method of enhancing NMR image contrast comprising: initiating a magnetic resonance device to image a sample, the magnetic resonance device generating a magnetization-dependent field; applying a pulse sequence to the magnetization-dependent field, wherein the pulse sequence places the magnetization-dependent field into an initial unstable configuration; evolving the magnetization-dependent field under the influence of a local feedback field to generate variations in a magnetization-dependent field distribution, wherein the variations in the magnetization-dependent field distribution operate on the magnetization- dependent field through the local feedback field to bring the magnetization-dependent field into a more stable state to amplify the contrast in the sample image in a positive feedback cycle; and imaging the stabilized magnetization-dependent field distribution using a spatially encoding detection sequence.

2. The method of claim 1, wherein the local feedback field is a field selected from the group consisting of a distant dipolar field, a radiation damping field and a joint reaction field, or a combination thereof.

3. The method of claim 2, wherein the distant dipolar field is utilized with samples having at least two contrasting regions with differences in one of either spin density or resonance frequency.

4. The method of claim 2, wherein the radiation damping field is utilized with samples having at least two contrasting regions with differing resonance frequencies.

5. The method of claim 1, wherein the magnetic resonance device wherein the contrast from the image arises from a magnetic resonance property selected from the group consisting of: spin density, resonance frequency, and proton density.

6. The method of claim 1, wherein the sample is selected from the group consisting of: dysplastic tissue, tumor tissue, necrotic tissue, cancerous tissue, and reversibly injured tissue.

7. The method of claim 1, further including preserving the sample in a 0.9% sodium chloride solution.

8. The method of claim 1, further comprising selecting a pulse sequence based on stability data for the magnetization-dependent fields under the local feedback fields.

9 The method of claim 1, wherein the NMR image contrast is based on one of the following parameters: the level of compartmentalized ferritin, the level of blood volume, the concentration of oxyhemoglobin, the concentration of deoxyhemoglobin, the concentration of methemoglobin, the blood oxygenation level, pH-dependent contrast, and amine-proton exchange rate variations.

10. The method of claim 1, further comprising disposing ferromagnetic nanoparticles in the vicinity of the sample.

11. A system for enhancing NMR image contrast comprising a circuit design to: generate a magnetization-dependent field directed into a sample to be imaged; apply a pulse sequence to the magnetization-dependent field, wherein the pulse sequence places the magnetization-dependent field into an initial unstable configuration; evolve the magnetization-dependent field under the influence of a local feedback field to generate variations in a magnetization-dependent field distribution, wherein the variations in the magnetization-dependent field distribution operate on the magnetization- dependent field through the local feedback field to bring the magnetization-dependent field into a more stable state to amplify the contrast in the sample image in a positive feedback cycle; and image the stabilized magnetization-dependent field distribution using a spatially encoding detection sequence.

12. The system of claim 11, wherein the local feedback field is a field selected from the group consisting of a distant dipolar field, a radiation damping field and a joint reaction field, or a combination thereof.

13. The system of claim 12, wherein the distant dipolar field is utilized with samples having at least two contrasting regions with differences in one of either spin density or resonance frequency.

14. The system of claim 12, wherein the radiation damping field is utilized with samples having at least two contrasting regions with differing resonance frequencies.

15. The system of claim 11, wherein the contrast from the image arises from a magnetic resonance property selected from the group consisting of: spin density, resonance frequency, and proton density.

16. The system of claim 11 wherein the sample is selected from the group consisting of: dysplastic tissue, tumor tissue, necrotic tissue, cancerous tissue, and reversibly injured tissue.

17. The system of claim 11, further comprising a circuit for selecting a pulse sequence based on stability data for the magnetization-dependent fields under the local feedback fields.

18 The system of claim 11, wherein the NMR image contrast is based on one of the following parameters: the level of compartmentalized ferritin, the level of blood volume, the concentration of oxyhemoglobin, the concentration of deoxyhemoglobin, the concentration of methemoglobin, the blood oxygenation level, pH-dependent contrast, and amine-proton exchange rate variations.

19. The system of claim 11, further comprising an injector for disposing ferromagnetic nanoparticles in the vicinity of the sample.