Dynamically adaptive MRI with encoding by singular value decomposition

A new MRI spatial encoding method based upon the singular value decomposition (SVD) and using spatially selective RF excitation is described. This encoding technique is particularly; applicable to dynamic adaptive MRI, because it provides a near minimal set of spatial encoding profiles computed using an image estimate that is determined from a previously obtained image. Experimental results are presented for two cases, which exemplify its potential use in different dynamic imaging tasks. SVD-encoded MRI has demonstrated to be a highly efficient encoding scheme.     

Implementation of a fast gradient-echo SVD encoding technique for dynamic imaging

In the paper, the results of a fast gradient-echo implementation of the singular value decomposition (SVD) encoding technique for dynamic imaging are presented. The method used is an adaptation with several critical modifications of a keyhole-type approach previously proposed but not implemented. The method was tested by imaging the events following injection of a contrast agent into a phantom, producing a series of dynamic image updates. It is demonstrated that, for this type of application, the SVD encoding technique adequately follows dynamic changes with even a small number of encodes. The result is comared qualitatively to that obtained by standard Fourier-based keyhole imaging and is shown to provide improved spatial resolution of dynamic events when updating with the same number of encodes.     

Cerebral blood flow index: dynamic perfusion MRI delivers a simple and good predictor for the outcome of acute-stage ischemic lesion

OBJECTIVES: To evaluate the feasibility of utilizing cerebral blood flow (CBF) index images, calculated automatically and quickly from dynamic perfusion imaging (DPI), to identify acute cerebral ischemia. We attempted to investigate (1) whether the CBF index has a threshold for assessing tissue outcome, (2) whether CBF index images can predict the resulting infracted area, and if so, (3) whether the predictive capacity of the CBF index image is comparable to the regional CBF (rCBF) image delivered from singular value decomposition (SVD) deconvolution methods, which are regarded as most accurate in predicting the final infarct area. METHODS: Diffusion-weighted images (DWI) and DPI were obtained in 17 patients within 12 hours of stroke onset and follow-up magnetic resonance imaging (MRI). On 3 DPI-delivered images, namely relative regional cerebral blood volume (rrCBV), uncorrected mean transit time (MTTu) and CBF index images, univariate discriminant analysis was done to estimate cut-off values to discriminate between infarcted and noninfarcted areas. Subsequently, correlations between the initial lesion volume of 3 images together with rCBF images delivered with SVD methods and the final infarct volume on follow-up T2-weighted MRI taken at the 8th to 20th day were determined. RESULTS: Among the 3 images, only the CBF index image was able reveal the threshold of the ischemic region. Lesion volume of CBF index images against follow-up infarct volume had the highest correlation (r = 0.995) to a linear fit and the slope was closest to 1.0 (0.91) among the 3 and had identical accuracy to the regression coefficient of rCBF images. CONCLUSIONS: CBF index images can predict final infarct volume. Evaluating CBF index images together with DWI can guide the initial assessment in the acute stage of cerebral ischemia.     

Dynamic range compression in MRI by means of a nonlinear gradient pulse

In current magnetic resonance imaging (MRI), valuable information must often be discarded because the NMR signal has greater dynamic range than the analog-to-digital converter (ADC) hardware. Typically, a small set of high-intensity data points near the center of the spin echo is responsible for most of the MRI data dynamic range. We predict that it is possible to reduce the dynamic range of the MRI spin echo by incorporating an identical nonlinear gradient pulse into each repetition of the imaging pulse sequence, prior to data sampling. This pulse converts the phase distribution of the subject, ordinarily a linear function of image coordinates, into a nonlinear function. A nonlinear phase distribution can have a negligible impact on image magnitude and yet a profound impact on spin-echo magnitude. Given a nonlinear phase distribution, there will no longer be a single data point at which all of the protons have an identical phase (the echo center). Instead, the protons become phase coherent on a piecemeal basis, the echo peak is smoothed out, and its maximum amplitude and dynamic range are greatly diminished. Using gradient pulses of quadratic spatial variation, we estimate that maximum echo amplitude and dynamic range can be reduced in most cases by an order of magnitude. © 1988 Academic Press, Inc.     

Effect of regional tracer delay on CBF in healthy subjects measured with dynamic susceptibility contrast-enhanced MRI: comparison with 15O-PET.

PURPOSE: Deconvolution based on truncated singular value decomposition (SVD deconvolution) is a promising method for measuring cerebral blood flow (CBF) with dynamic susceptibility contrast-enhanced magnetic resonance imaging (DSC-MRI), but it has proved extremely sensitive to tracer delay. The purpose of this study was to investigate the effect of regional tracer delay on CBF determined by SVD deconvolution (SVD-CBF). SVD-CBFs with and without correction for the delay were compared with CBF measured by positron emission tomography (PET-CBF), which is regarded as the gold standard for quantification of CBF. METHODS: Perfusion MRI and PET were performed on seven healthy men. In the PET study, the CBF image was obtained with bolus injection of H2(15)O and continuous arterial sampling. In the DSC-MRI study with bolus injection of Gd-based contrast agent, dynamic perfusion data were obtained with a 1.5T scanner at 1-s intervals by means of gradient-echo echo-planar imaging. CBF was determined by the SVD deconvolution method with and without correction for the tracer delay. Region-of-interest measurements were obtained in the gray matter (cerebral cortex in the middle cerebral artery territory) and white matter (centrum semiovale). RESULTS: Tracer delay was significantly longer in white matter than in gray matter (1.45+/-0.61 s vs. 0.59+/-0.35 s, P<0.01). Correction for the delay increased SVD-CBF in the white matter and consequently reduced the gray-to-white SVD-CBF ratio. The uncorrected gray-to-white SVD-CBF ratio was significantly larger than that of PET-CBF (3.33+/-0.66 vs. 2.54+/-0.49, P<0.01). However, the gray-to-white delay-corrected SVD-CBF ratio did not differ significantly from that of PET-CBF (2.83+/-0.31 vs. 2.54+/-0.49, P=0.10). CONCLUSION: The tracer delay in DSC-MRI causes errors in CBF estimates, even in healthy persons, and therefore should be corrected for when delay-sensitive deconvolution, such as SVD deconvolution, is used.     

Analysis of scintigrams by singular value decomposition (SVD) technique

The singular value decomposition (SVD) method is presented as a potential tool for analyzing gamma camera images. Mathematically image analysis is a study of matrixes as the standard scintigram is a digitized matrix presentation of the recorded photon fluence from radioactivity of the object. Each matrix element (pixel) consists of a number, which equals the detected counts of the object position. The analysis of images can be reduced to the analysis of the singular values of the matrix decomposition. In the present study the clinical usefulness of SVD was tested by analyzing two different kinds of scintigrams: brain images by single photon emission tomography (SPET), and liver and spleen planar images. It is concluded that SVD can be applied to the analysis of gamma camera images, and that it provides an objective method for interpretation of clinically relevant information contained in the images. In image filtering, SVD provides results comparable to conventional filtering. In addition, the study of singular values can be used for semiquantitation of radionuclide images as exemplified by brain SPET studies and liver-spleen planar studies.     

TrueFisp versus HASTE sequences in 3T cine MRI: Evaluation of image quality during phonation in patients with velopharyngeal insufficiency

Objective

To evaluate the image quality of two fast dynamic magnetic resonance imaging (MRI) sequences: True fast imaging with steady state precession (TrueFisp) was compared with half-Fourier acquired single turbo-spin-echo (HASTE) sequence for the characterization of velopharyngeal insufficiency (VPI) in repaired cleft palate patients.

Methods

Twenty-two patients (10 female and 12 male; mean age, 17.7?±?10.6 years; range, 9–31) with suspected VPI underwent 3-T MRI using TrueFisp and HASTE sequences. Imaging was performed in the sagittal plane at rest and during phonation of “ee” and “k” to assess the velum, tongue, posterior pharyngeal wall and a potential VP closure. The results were analysed independently by one radiologist and one orthodontist.

Results

HASTE performed better than TrueFisp for all evaluated items, except the tongue evaluation by the orthodontist during phonation of “k” and “ee”. A statistically significant difference in favour of HASTE was observed in assessing the velum at rest and during phonation of “k” and “ee”, and also in assessing VP closure in both raters (p?<?0.05). TrueFisp imaging was twice as fast as HASTE (0.36 vs. 0.75 s/image).

Conclusion

Dynamic HASTE images were of superior quality to those obtained with TrueFisp, although TrueFisp imaging was twice as fast.

Key Points

? Dynamic MRI is an invaluable tool for diagnosing VPI. ? Dynamic HASTE images were of superior quality to those obtained with TrueFisp. ? TrueFisp imaging was twice as fast as HASTE imaging.

     

Simulation of spatial and contrast distortions in keyhole imaging

Keyhole imaging is a scheme introduced to improve temporal resolution in dynamic contrast-enhanced MRI by a factor of four or more. A “full” acquisition before contrast administration is followed by truncated acquisitions sensitive primarily to changes in image contrast. Simulations of the point-spread functions that obtain, and their effect on contrast and spatial resolution, reveal significant degradation only for the smallest objects. Our simulations also address the feasibility of three-dimensional keyhole imaging, and demonstrate a potential 16-fold increase in temporal resolution. This suggests roles for keyhole imaging in conventional (nondynamic) precon-trast and postcontrast studies and other applications.     

Adaptive keyhole methods for dynamic magnetic resonance image reconstruction.

Dynamic magnetic resonance imaging (MRI) acquires a sequence of images for the visualization of the temporal variation of tissue or organs. Keyhole methods such as Fourier keyhole (FK) and keyhole SVD (KSVD) are the most popular methods for image reconstruction in dynamic MRI. This paper provides a class of adaptive keyhole methods, called adaptive FK (AFK) and adaptive KSVD (AKSVD), for dynamic MRI reconstruction. The proposed methods are based on the conventional Fourier encoding and SVD encoding schemes. Instead of the conventional keyhole methods' duplication of un-acquired data from the reference images, the proposed methods use a temporal model to depict the inter-frame dynamic changes and to estimate the un-acquired data in each successive frame. Because the model is online identified from the acquired data, the proposed methods do not require the pre-imaging process, the navigator signals, and any prior knowledge of the imaged objects. Furthermore, the new methods use the conventional keyhole encoding schemes without the bias to any particular object characters, and the temporal model for updating information is in the general form of AR process without the preference to any particular motion types. Hence, the proposed methods are designed as a generic approach to dynamic MRI, other than for any specific class of objects. Studies on dynamic MRI data set show that the new methods can produce images with much lower reconstruction error than the conventional FK and KSVD.     

Interactive scan control for kinematic study in open MRI

PURPOSE: A tool to support the subject is generally used for kinematic joint imaging with an open MRI apparatus because of difficulty setting the image plane correctly. However, use of a support tool requires a complicated procedure to position the subject, and setting the image plane when the joint angle changes is time consuming. Allowing the subject to move freely enables better diagnoses when kinematic joint imaging is performed. We therefore developed an interactive scan control (ISC) to facilitate the easy, quick, and accurate setting of the image plane even when a support tool is not used. METHODS: We used a 0.4T magnetic resonance (MR) imaging system open in the horizontal direction. The ISC determines the image plane interactively on the basis of fluoroscopy images displayed on a user interface. The imaging pulse is a balanced steady-state acquisition with rewound gradient echo (SARGE) sequence with update time less than 2 s. Without using a tool to support the knee, we positioned the knee of a healthy volunteer at 4 different joint angles and set the image plane through the patella and femur at each of the angles. Lumbar imaging is also demonstrated with ISC. RESULTS: Setting the image plane was easy and quick at all knee angles, and images obtained clearly showed the patella and femur. Total imaging time was less than 10 min, a fourth of the time needed when a support tool is used. We also used our ISC in kinematic imaging of the lumbar. CONCLUSION: The ISC shortens total time for kinematic joint imaging, and because a support tool is not needed, imaging can be done more freely in an open MR imaging apparatus.     

Adaptive bilateral breast MRI using projection reconstruction time-resolved imaging of contrast kinetics

PURPOSE: To introduce a bilateral implementation of an adaptive imaging technique in which both dynamic and high resolution breast MR images are acquired simultaneously. MATERIALS AND METHODS: Adaptive three-dimensional bilateral breast imaging in the sagittal plane was achieved by combining two elements: a projection reconstruction time-resolved imaging of contrast kinetics (PR-TRICKS) k-space trajectory and a slab interleaved sequence that imaged alternate breasts every TR. A pilot study was performed to evaluate image quality and contrast uptake behavior, using eight patients with previously identified benign lesions. RESULTS: Adaptive reconstruction demonstrated breast lesions in all eight women with similar image quality and signal-to-noise ratio (SNR) to Cartesian images with comparable imaging parameters. Contrast enhancement curves covering the entire postinjection time period were obtained from the dynamic images and in one case compared to previous enhancement profiles from a conventional Cartesian trajectory. CONCLUSION: Bilateral dynamic and high spatial resolution images with high SNR can be achieved in a clinically feasible manner, providing both kinetic and morphologic analysis with a single data set. This may obviate the need for multiple MRI examinations for a thorough breast MRI workup.     

Effect of vascular stenosis on perfusion-weighted imaging; differences between calculation algorithms

PURPOSE: To determine the most suitable postprocessing technique for magnetic resonance (MR) perfusion imaging in patients with vascular stenosis, by comparing the cerebral blood flow (CBF) maps of single photon emission tomography (SPECT) and perfusion MR imaging (MRI). MATERIALS AND METHODS: In 15 consecutive patients (14 men and one woman, mean age 73.9 +/- 6.0 years) with stenosis of common carotid artery (CCA) or internal carotid artery (ICA) of more than 75%, both brain perfusion MRI and brain perfusion SPECT were performed. From perfusion MR images, CBF maps were calculated with the first moment, singular value decomposition (SVD), and block circulant SVD (b-SVD) methods, and CBF maps from each algorithm were compared with those from SPECT. RESULTS: The b-SVD method had the best correlation with SPECT (R = 0.814), followed by the first moment method (R = 0.776) and the SVD method (R = 0.723). The b-SVD method has the least mean difference with SPECT (0.118), the first moment method also had less difference (0.121), and the SVD had greatest mean difference (0.164). CONCLUSION: Our results suggest that in patients with vascular impairment the b-SVD method will be the technique of choice rather than SVD or first moment method.     

Angle‐sensitive MRI for quantitative analysis of fiber‐network deformations in compressed cartilage

Purpose

To demonstrate the feasibility of a novel experimental method to quantitatively analyze fiber‐network deformation in compressed cartilage by angle‐sensitive magnetic resonance imaging (MRI) of cartilage.

Methods

Three knee cartilage samples of an adult sheep were imaged in a high‐resolution MRI scanner at 7 T. Main fiber orientation and its “offset” from the direction perpendicular to the bone‐cartilage boundary were derived from MR images taken at different orientations with respect to B₀. Bending of the collagen fibers was determined from weight‐bearing MRI with the load (up to 1.0 MPa) applied over the whole sample surface. A “fascicle” model of the cartilage ultrastructure was assumed to analyze characteristic intensity variations in T₂‐weighted images under load.

Results

T₂‐weighted MR images showed a strong variation of the signal intensities with sample orientation. In the T₂‐weighted weight‐bearing series, regions of high signal intensity underwent shifts from the lateral to the central parts in all three cartilage samples. The bending of the collagen fibers was determined to be 27.2°, 35.4°, and 40.0° per MPa, respectively.

Conclusion

Assuming a “fascicle” model, the presented MRI method provides quantitative measures of structural adjustments in compressed cartilage. Our preliminary analysis suggests that cartilage fiber deformation includes both bending and crimping.     

A simulation study to assess SVD encoding for interventional MRI: Effect of object rotation and needle insertion

Singular value decomposition (SVD) encoding offers great promise to provide high spatial and temporal resolution required for interventional MRI (I-MRI) (1). This study investigates its efficacy when (a) objects are rotated and (b) a small device (ie, a needle) is moved within anatomic structures. It was found that SVD-encoded MRI is biased toward the reference from which encoding vectors are derived, thus providing a potential limitation under conditions in which the object has undergone significant global change. Reference images with partial device insertion may be needed to accurately resolve the device or track the object motion. Theoretically, the differences between the reference and the object being imaged suggest that SVD encoding is suboptimal (in a minimum mean squared error sense). Other encoding/reconstruction algorithms may come closer to achieving the desired advantages in spatial and temporal fidelity.     

Comparison of dynamic MRI at 3.0T and MDCT of pancreatobiliary disease: Evaluation with source,MPR, CPR,and MIP images for image quality and hepatic arterial and portal venous vessel conspicuity

Purpose

To compare contrast material‐enhanced three‐dimensional (3D) magnetic resonance imaging (MRI) at 3.0T and multidetector row computed tomography (MDCT) in the same patient with regard to image quality of pancreatobiliary disease and hepatic vascular conspicuity.

Materials and Methods

This study enrolled 32 patients with pancreatobiliary disease who underwent both gadolinium‐enhanced 3D dynamic MRI and multiphasic CT using 16‐MDCT. Data analysis of image quality was performed by two radiologists based on source images, multiplanar reconstruction (MPR), curved planar reconstruction (CPR), and maximum intensity projection (MIP) reconstruction. Determination of image quality was based on a 4‐point image quality rating scale.

Results

The overall image quality of the MRI axial images was superior to that of the axial MDCT images. The MRI protocol yielded an average score of 3.8 points versus 3.5 for the CT imaging. No significant difference was found between 3.0T MRI and MDCT images in MPR or CPR image quality. Image quality for visualization of the distal intrahepatic segmental arteries was significantly improved using MDCT imaging. No significant difference was found between the MDCT and 3.0T MR in portal vein branch image quality.

Conclusion

High‐resolution dynamic contrast‐enhanced MR imaging at 3.0T is a comprehensive technique which provides high image quality in pancreatobiliary disease. J. Magn. Reson. Imaging 2009;29:846–852. © 2009 Wiley‐Liss, Inc.     

Functional MRI using regularized parallel imaging acquisition.

Parallel MRI techniques reconstruct full-FOV images from undersampled k-space data by using the uncorrelated information from RF array coil elements. One disadvantage of parallel MRI is that the image signal-to-noise ratio (SNR) is degraded because of the reduced data samples and the spatially correlated nature of multiple RF receivers. Regularization has been proposed to mitigate the SNR loss originating due to the latter reason. Since it is necessary to utilize static prior to regularization, the dynamic contrast-to-noise ratio (CNR) in parallel MRI will be affected. In this paper we investigate the CNR of regularized sensitivity encoding (SENSE) acquisitions. We propose to implement regularized parallel MRI acquisitions in functional MRI (fMRI) experiments by incorporating the prior from combined segmented echo-planar imaging (EPI) acquisition into SENSE reconstructions. We investigated the impact of regularization on the CNR by performing parametric simulations at various BOLD contrasts, acceleration rates, and sizes of the active brain areas. As quantified by receiver operating characteristic (ROC) analysis, the simulations suggest that the detection power of SENSE fMRI can be improved by regularized reconstructions, compared to unregularized reconstructions. Human motor and visual fMRI data acquired at different field strengths and array coils also demonstrate that regularized SENSE improves the detection of functionally active brain regions.     

Applicability and efficiency of near-optimal spatial encoding for dynamically adaptive MRI

Adaptive near-optimal MRI spatial encoding entails, for the acquisition of each image update in a dynamic series, the computation of encodes in the form of a linear algebra-derived orthogonal basis set determined from an image estimate. The origins of adaptive encoding relevant to MRI are reviewed. Sources of error of this approach are identified from the linear algebraic perspective where MRI data acquisition is viewed as the projection of information from the field-of-view onto the encoding basis set. The definitions of ideal and non-ideal encoding follow, with nonideal encoding characterized by the principal angles between two vector spaces. An analysis of the distribution of principal angles is introduced and applied in several example cases to quantitatively describe the suitability of a basis set derived from a specific image estimate for the spatial encoding of a given field-of-view. The robustness of adaptive near-optimal spatial encoding for dynamic MRI is favorably shown by results computed using singular value decomposition encoding that simulates specific instances of worst case data acquisition when all objects have changed or new objects have appeared in the field-of-view. The mathematical analysis and simulations presented clarify the applicability and efficiency of adaptively determined near-optimal spatial encoding throughout a range of circumstances as may typically occur during use of dynamic MRI.     

Fast dixon‐based multisequence and multiplanar MRI for whole‐body detection of cancer metastases

Purpose

To develop and demonstrate the feasibility of multisequence and multiplanar MRI for whole‐body cancer detection.

Materials and Methods

Two fast Dixon‐based sequences and a diffusion‐weighted sequence were used on a commercially available 1.5 T scanner for whole‐body cancer detection. The study enrolled 19 breast cancer patients with known metastases and in multistations acquired whole‐body axial diffusion‐weighted, coronal T2‐weighted, axial/sagittal pre‐ and postcontrast T1‐weighted, as well as triphasic abdomen images. Three radiologists subjectively scored Dixon images of each series for overall image quality and fat suppression uniformity on a 4‐point scale (1 = poor, 2 = fair, 3 = good, and 4 = excellent).

Results

Eighteen of the 19 patients completed the whole‐body MRI successfully. The mean acquisition time and overall patient table time were 46 ± 3 and 69 ± 5 minutes, respectively. The average radiologists' scores for overall image quality and fat suppression uniformity were both 3.4 ± 0.5. The image quality was consistent between patients and all completed whole‐body examinations were diagnostically adequate.

Conclusion

Whole‐body MRI offering essentially all the most optimal tumor‐imaging sequences in a typical 1‐hour time slot can potentially become an appealing “one‐stop‐shop” for whole‐body cancer imaging. J. Magn. Reson. Imaging 2009;29:1154–1162. © 2009 Wiley‐Liss, Inc.     

Registration of [18F]FDG microPET and small-animal MRI

This paper describes a voxel-based method for coregistering microPET [(18)F]FDG emission images and MRI data without the need for fiducial markers. [(18)F]FDG has a well-characterized biodistribution in normal mice and thus may be useful for image registration. Female BALB/c mice were implanted with EMT-6 mouse mammary carcinoma 1 week prior to imaging. Three imaging sessions were performed in which a [(18)F]FDG microPET-R4 emission scan was taken followed by small-animal MRI with and without Gd-based contrast agent. MicroPET and MR images were registered using a voxel-based algorithm that computes rigid-body image transformations based on the alignment of intensity gradients. Registration accuracy was assessed on the basis of dual-modality external fiducial line sources incorporated into the mouse bed. The root mean square (rms) registration errors were 0.74 mm translation and 1.44 degrees rotation without contrast and 0.72 mm translation and 0.89 degrees rotation with contrast. Generally, good registration was evident upon inspection of fused microPET/MR images. Accurate automated, voxel-based microPET-MR image coregistration, utilizing image intensity gradients, is feasible. Our technique requires no manual identification of image features and makes no use of surgically implanted or external fiducial markers or stereotactic apparatus.     

Contrast-enhanced ultra-high-field liver MRI: A feasibility trial

The aim of this study was to investigate the feasibility of dynamic contrast-enhanced 7 T MRI of the liver using an eight-channel radiofrequency (RF) transmit/receive body-coil. 16 healthy subjects were examined on a 7 T MR system utilizing a custom-built eight-channel RF body-coil suitable for RF-shimming. The following data were acquired: (1) steady state free precession imaging, (2) T2w turbo spin echo imaging, (3) T1w in and opposed-phase imaging, (4) T1w 3D FLASH images pre-contrast and in arterial, portal-venous and venous phase and (5) a fat-saturated pre- and post-contrast 2D FLASH sequence. Visual evaluation of (1) the delineation of liver vasculature, (2) the overall image quality, and (3) artifact presence and consequent image impairment was performed. SNR of the liver parenchyma was measured for the contrast-enhanced 2D and 3D FLASH sequences. For statistical analysis, a Wilcoxon-Rank Test was used. Best delineation of non-enhanced liver vasculature and overall image quality was found for 2D FLASH MRI, with only slight improvement in vessel conspicuity after the application of contrast media. T2-weighted TSE imaging remained strongly impaired, falling short of diagnostic relevance and precluding a clinical application. Our results demonstrate the feasibility and diagnostic potential of dedicated contrast-enhanced 7 T liver MRI as well as the potential for non-contrast-enhanced angiographic application.