The relationship between brain structure and complex behavior is governed by large-scale neurocognitive networks. The availability of a noninvasive technique that can visualize the neuronal projections connecting the functional centers should therefore provide new keys to the understanding of brain function. By using high-resolution three-dimensional diffusion magnetic resonance imaging and a newly designed tracking approach, we show that neuronal pathways in the rat brain can be probed in situ. The results are validated through comparison with known anatomical locations of such fibers.

Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging.

A magnetic resonance (MR) imaging apparatus and technique exploits spatial information inherent in a surface coil array to increase MR image acquisition speed, resolution and/or field of view. Partial signals are acquired simultaneously in the component coils of the array and formed into two or more signals corresponding to orthogonal spatial representations. In a Fourier embodiment, lines of the k-space matrix required for image production are formed using a set of separate, preferably linear combinations of the component coil signals to substitute for spatial modulations normally produced by phase encoding gradients. The signal combining may proceed in a parallel or flow-through fashion, or as post-processing, which in either case reduces the need for time-consuming gradient switching and expensive fast magnet arrangements. In the post-processing approach, stored signals are combined after the fact to yield the full data matrix. In the flow-through approach, a plug-in unit consisting of a coil array with an on board processor outputs two or more sets of combined spatial signals for each spin conditioning cycle, each directly corresponding to a distinct line in k-space. This partially parallel imaging strategy, dubbed SiMultaneous Acquisition of Spatial Harmonics (SMASH), is readily integrated with many existing fast imaging sequences, yielding multiplicative time savings without a significant sacrifice in spatial resolution or signal-to-noise ratio. An experimental system achieved two-fold improvement in image acquisition time with a prototype three-coil array, and larger factors are achievable with ther coil arrangements.

Simultaneous acquisition of spatial harmonics (SMASH): ultra-fast imaging with radiofrequency coil arrays

Magnetic resonance imaging apparatus

Compressed sensing magnetic resonance imaging (CS-MRI) enables fast acquisition, which is highly desirable for numerous clinical applications. This can not only reduce the scanning cost and ease patient burden, but also potentially reduce motion artefacts and the effect of contrast washout, thus yielding better image quality. Different from parallel imaging-based fast MRI, which utilizes multiple coils to simultaneously receive MR signals, CS-MRI breaks the Nyquist–Shannon sampling barrier to reconstruct MRI images with much less required raw data. This paper provides a deep learning-based strategy for reconstruction of CS-MRI, and bridges a substantial gap between conventional non-learning methods working only on data from a single image, and prior knowledge from large training data sets. In particular, a novel conditional Generative Adversarial Networks-based model (DAGAN)-based model is proposed to reconstruct CS-MRI. In our DAGAN architecture, we have designed a refinement learning method to stabilize our U-Net based generator, which provides an end-to-end network to reduce aliasing artefacts. To better preserve texture and edges in the reconstruction, we have coupled the adversarial loss with an innovative content loss. In addition, we incorporate frequency-domain information to enforce similarity in both the image and frequency domains. We have performed comprehensive comparison studies with both conventional CS-MRI reconstruction methods and newly investigated deep learning approaches. Compared with these methods, our DAGAN method provides superior reconstruction with preserved perceptual image details. Furthermore, each image is reconstructed in about 5 ms, which is suitable for real-time processing.

/pdf/dagan-deep-de-aliasing-generative-adversarial-networks-for-3u7cltdv1j.pdf

DAGAN: Deep De-Aliasing Generative Adversarial Networks for Fast Compressed Sensing MRI Reconstruction

Snapshot FLASH magnetic resonance imaging techniques have been developed to enable real-time imaging of MR parameters. The first realization of the method is based on a 64 × 128 FLASH tomogram acquired within 200 ms, using improved MR system hardware conditions. The soft tissue contrast obtained in FLASH MRI almost disappears by using flip angles of less than 5° and repetition times of 3 ms. This work describes extensions of FLASH MRI placing conventional MR experiments before the whole imaging sequence. This creates images of any desired contrast without changing the measuring time. Examples of inversion-recovery T1, spin-echo T2, chemical-shift-selective, and spectroscopic FLASH MRI are presented. Further extensions to real-time MRI of blood vessels, diffusion coefficients, combination with two-dimensional MR spectroscopy experiments, and other nuclei are discussed.

Snapshot FLASH MRI. Applications to T1, T2, and chemical-shift imaging.

Based on the principles of echo imaging, we present a method to acquire sufficient data for a 256 X 256 image in from 2 to 40 s. The image contrast is dominated by the transverse relaxation time T2. Sampling all projections for 2D FT image reconstruction in one (or a few) echo trains leads to image artifacts due to the different T2 weighting of the echo. These artifacts cannot be described by a simple smearing out of the image in the phase direction. Proper distribution of the phase-encoding steps on the echoes can be used to minimize artifacts and even lead to resolution enhancement. In spite of the short data acquisition times, the signal amplitudes of structures with long T2 are nearly the same as those in a conventional 2D FT experiment. Our method, therefore, is an ideal screening technique for lesions with long T2.

RARE imaging: a fast imaging method for clinical MR.

The RARE method is based on the principle of echo imaging to generate images with high T2 contrast. Since RARE is a fast imaging method, it can be used to acquire a high-resolution three-dimensional data set in less than 15 min. Thin slices from such a three-dimensional data set provide detailed information about the ventricular system. © 1987 Academic Press, Inc.

Fast three-dimensional imaging of cerebrospinal fluid.

It is demonstrated that it is possible to acquire two volume selective 1H NMR spectra of human brain in vivo, consisting of voxels of 1.5 × 1.5 × 1.5 cm3, within 14 min with a good S/N ratio. This is mainly achieved by the application of a PRESS sequence generating a spin-echo of the VOI at 135 ms in conjunction with the STABLE technique by which two spectra can be recorded in an interlaced mode. The Bo homogeneity over such small voxels is considerably higher than over larger voxels. With these methodological improvements it is possible to observe morphological heterogeneity of tumors. The results indicate that spectral changes seem to correlate with the metabolic state of the tumor rather than the tumor type. Additionally the spectrum of a patient with multiple sclerosis suggests that even differentiation between tumors and other lesions might not be possible.

H. Friedburg

Papers

RARE imaging: a fast imaging method for clinical MR.

Fast three-dimensional imaging of cerebrospinal fluid.

The importance of the voxel size in clinical 1H spectroscopy of the human brain.

Measurement of CSF flow using an interferrographic MR technique based on the RARE-fast imaging sequence