Application of keyhole imaging to interventional MRI: A simulation study to predict sequence requirements

A wide variety of techniques have been proposed recently to improve the temporal resolution of MRI. These include echo-planar imaging methods, wavelet encoding, singular value decomposition encoding, and k-space sharing methods known as ?keyhole”? imaging. In this work, we use a simulation study to investigate the phase-encoding ordering and data-sharing methods required for the application of keyhole imaging to interventional MRI (I-MRI). The advantages of keyhole imaging over other methods are its simplicity and the use of conventional phase encoding and Fourier transform reconstruction found on virtually all modern MR imagers. Our analysis has predicted that conventional keyhole methods that repeatedly acquire only the center portion of k space, and those that sequentially progress from the center of k space outward, will not meet the combination of temporal and spatial resolution required for tip localization during I-MRI needle insertion. Instead, acquisitions that acquire both high and low k-space data, in ranked order, should provide acceptable tip position and needle width accuracy in both temporal and spatial domains for use in I-MRI.     

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.     

Improving the temporal resolution of functional MR imaging using keyhole techniques

Using a keyhole technique, it is shown that the data acquisition rate of gradient-echo imaging for functional MRI (fMRI) studies can be increased substantially. The resulting enhancement of the temporal resolution of fMRIs was accomplished without modifying the hardware of a conventional MRI system. High spatial resolution fMRI images were first collected with conventional full k-space acquisition and image reconstruction. Using the same data set, simulation reconstruction using the keyhole principle and zero-padding were performed for comparison with the full k-space reconstruction. No significant changes were found for fMRI images generated from the keyhole technique with a data sharing profile of 50% of the k-space. As k-space data sharing profiles increased to 75 and 87.5%, the keyhole fMRI images began to show only modest changes in activation intensity and area compared with the standard images. In contrast, zero-padding fMRI images produced a significant disparity both in activation intensity and area relative to the truly high-resolution fMRI images. The keyhole technique's ability to retain the intensity and area of fMRI information, while substantially reducing acquisition time, makes it a promising method for fMRI studies.     

On the Relationship Between Feature-Recognizing MRI and MRI Encoded by Singular Value Decomposition

This paper describes the similarity between two methods of non-Fourier MRI: feature-recognizing MRI (FR MRI) and MRI with encoding by singular value decomposition (SVD MRI). Both methods represented images as truncated expansions of non-Fourier basis functions; these basis images were derived from prior image data by using closely-related mathematical techniques: the Karhunen-Loeve decomposition (or principal components analysis) and singular value decomposition, respectively. We demonstrate that FR and SVD MRI are equivalent in the following sense: given the same prior image data, they lead to exactly the same basis functions. FR MRI utilized prior images of the same body part in many “training” subjects, thought to be similar to the “unknown” subject to be imaged. SVD MRI utilized a single prior image of one subject in order to perform dynamic imaging of that subject. We demonstrate that the basis function expansion derived from a single prior image may not be capable of representing new features (features not found in the prior image). Therefore, the SVD basis functions may be inappropriate for dynamic imaging.     

A comparison of RIGR and SVD dynamic imaging methods

Several constrained imaging methods have recently been proposed for dynamic imaging applications. This paper compares two of these methods: the Reduced-encoding Imaging by Generalized-series Reconstruction (RIGR) and Singular Value Decomposition (SVD) methods. RIGR utilizes a priori data for optimal image reconstruction whereas the SVD method seeks to optimize data acquisition. However, this study shows that the existing SVD encoding method tends to bias the data acquisition scheme toward reproducing the known features in the reference image. This characteristic of the SVD encoding method reduces its capability to capture new image features and makes it less suitable than RIGR for dynamic imaging applications.     

Improving k-t SENSE by adaptive regularization.

The recently proposed method known as k-t sensitivity encoding (SENSE) has emerged as an effective means of improving imaging speed for several dynamic imaging applications. However, k-t SENSE uses temporally averaged data as a regularization term for image reconstruction. This may not only compromise temporal resolution, it may also make some of the temporal frequency components irrecoverable. To address that issue, we present a new method called spatiotemporal domain-based unaliasing employing sensitivity encoding and adaptive regularization (SPEAR). Specifically, SPEAR provides an improvement over k-t SENSE by generating adaptive regularization images. It also uses a variable-density (VD), sequentially interleaved k-t space sampling pattern with reference frames for data acquisition. Simulations based on experimental data were performed to compare SPEAR, k-t SENSE, and several other related methods, and the results showed that SPEAR can provide higher temporal resolution with significantly reduced image artifacts. Ungated 3D cardiac imaging experiments were also carried out to test the effectiveness of SPEAR, and real-time 3D short-axis images of the human heart were produced at 5.5 frames/s temporal resolution and 2.4 x 1.2 x 8 mm3 spatial resolution with eight slices.     

Patient‐adaptive reconstruction and acquisition in dynamic imaging with sensitivity encoding (PARADISE)

MRI of the human heart without explicit cardiac synchronization promises to extend the applicability of cardiac MR to a larger patient population and potentially expand its diagnostic capabilities. However, conventional nongated imaging techniques typically suffer from low image quality or inadequate spatio‐temporal resolution and fidelity. Patient‐Adaptive Reconstruction and Acquisition in Dynamic Imaging with Sensitivity Encoding (PARADISE) is a highly accelerated nongated dynamic imaging method that enables artifact‐free imaging with high spatio‐temporal resolutions by utilizing novel computational techniques to optimize the imaging process. In addition to using parallel imaging, the method gains acceleration from a physiologically driven spatio‐temporal support model; hence, it is doubly accelerated. The support model is patient adaptive, i.e., its geometry depends on dynamics of the imaged slice, e.g., subject's heart rate and heart location within the slice. The proposed method is also doubly adaptive as it adapts both the acquisition and reconstruction schemes. Based on the theory of time‐sequential sampling, the proposed framework explicitly accounts for speed limitations of gradient encoding and provides performance guarantees on achievable image quality. The presented in‐vivo results demonstrate the effectiveness and feasibility of the PARADISE method for high‐resolution nongated cardiac MRI during short breath‐hold. Magn Reson Med, 2010. © 2010 Wiley‐Liss, Inc.     

Improved device definition in interventional magnetic resonance imaging using a rotated stripes keyhole acquisition.

Keyhole acquisition techniques have been used to reduce image acquisition times primarily in contrast agent studies and via simulation in interventional MRI procedures. More recent simulations have suggested that improved definition of an interventional device [e.g., biopsy needles, radio frequency (RF) electrodes] could be achieved by rotating the keyhole pattern in k-space so that the read out direction lies perpendicular to the device orientation in real space. This study seeks to validate the earlier predictions of improved efficacy of a rotated stripes keyhole acquisition in actual in vitro and in vivo interventional MR imaging procedures. A true-FISP sequence was modified to perform central stripes keyhole (as known as conventional keyhole) acquisitions after a full initial reference data set was acquired. The gradients of this sequence were then modified to rotate the k-space definition and the keyhole stripes by 10 degrees, 20 degrees, 30 degrees, 45 degrees, and 60 degrees from their conventional k-space orientation. Acquisitions were performed during insertion of interventional devices in phantom and in vivo RF ablation procedures, using the modified sequence selected which placed the phase encoding axis at parallel and perpendicular orientations to the devices. Resulting images were compared between the two orientations for needle width and tip accuracy. Apparent needle width was thinner and tip position more accurately determined for placement of phase encoding parallel to the needle in all cases. Rotated keyhole imaging provides the required temporal advantage of conventional keyhole imaging along with a near optimal definition of an interventional device when the phase encoding is oriented parallel to the direction of the needle motion. Magn Reson Med 42:554-560, 1999.     

Dynamic MRI with projection reconstruction and KWIC processing for simultaneous high spatial and temporal resolution.

A method for dynamic imaging in MRI is presented that enables the acquisition of a series of images with both high temporal and high spatial resolution. The technique, which is based on the projection reconstruction (PR) imaging scheme, utilizes distinct data acquisition and reconstruction strategies to achieve this simultaneous capability. First, during acquisition, data are collected in multiple undersampled passes, with the view angles interleaved in such a way that those of subsequent passes bisect the views of earlier ones. During reconstruction, these views are weighted according to a previously described k-space weighted image contrast (KWIC) technique that enables the manipulation of image contrast by selective filtering. Unlike conventional undersampled PR methods, the proposed dynamic KWIC technique does not suffer from low image SNR or image degradation due to streaking artifacts. The effectiveness of dynamic KWIC is demonstrated in both simulations and in vivo, high-resolution, contrast-enhanced imaging of breast lesions.     

Radial single-shot STEAM MRI.

Rapid MR imaging using the stimulated echo acquisition mode (STEAM) technique yields single-shot images without any sensitivity to resonance offset effects. However, the absence of susceptibility-induced signal voids or geometric distortions is at the expense of a somewhat lower signal-to-noise ratio than EPI. As a consequence, the achievable spatial resolution is limited when using conventional Fourier encoding. To overcome the problem, this study combined single-shot STEAM MRI with radial encoding. This approach exploits the efficient undersampling properties of radial trajectories with use of a previously developed iterative image reconstruction method that compensates for the incomplete data by incorporating a priori knowledge. Experimental results for a phantom and human brain in vivo demonstrate that radial single-shot STEAM MRI may exceed the resolution obtainable by a comparable Cartesian acquisition by a factor of four.     

Shared k-space echo planar imaging with keyhole.

Time-dependent phenomena are of great interest, and researchers have sought to shed light on these processes with MRI, particularly in vivo. In this work, a new hybrid technique based on EPI and using the concept of keyhole imaging is presented. By sharing peripheral k-space data between images and acquiring the keyhole more frequently, it is shown that the spatial resolution of the reconstructed images can be maintained. The method affords a higher temporal resolution and is more robust against susceptibility and chemical-shift artifacts than single-shot EPI. The method, termed shared k-space echo planar imaging with keyhole (shared EPIK), has been implemented on a standard clinical scanner. Technical details, simulation results, phantom images, in vivo images, and fMRI results are presented. These results indicate that the new method is robust and may be used for dynamic MRI applications. Magn Reson Med 45:109-117, 2001.     

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.     

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.     

Excitation UNFOLD (XUNFOLD) to improve the temporal resolution of multishot tailored RF pulses.

An extension of the "UNaliasing by Fourier encoding the Overlaps using the temporaL Dimension" (UNFOLD) method to the excitation domain (XUNFOLD) is presented to improve the temporal resolution of multishot tailored RF (TRF) pulses. Multishot three-dimensional TRF pulses were designed to produce a time series of images with periodically aliased excitation profiles. The XUNFOLD method is shown to remove the excitation profile aliasing from the dynamic imaging data by filtering in the temporal frequency dimension. The technique is demonstrated to improve the temporal resolution of simulated functional MRI (fMRI) activation in a time series of brain images.     

Does the "keyhole" technique improve spatial resolution in MRI perfusion measurements? A study in volunteers

We examined the potential of the ’keyhole' technique to improve spatial resolution in perfusion-weighted MRI on whole-body imagers with standard gradient hardware. We examined 15 healthy volunteers. We acquired a high-resolution image with 256 phase-encoding steps before a bolus-tracking procedure. For the dynamic series we collected only 34 lines in the center of k-space. Data reconstruction was performed by both zero-filling and keyhole methods. The dynamic datasets, concentration-time curves calculated from user-defined regions and maps of the cerebrovascular parameters using both reconstruction methods were compared. Using keyhole series, anatomical structures could easily be defined which were not seen on the original dynamic series because of blurring due to ringing artefacts. Comparison of signal-time curves in large regions yielded no significant difference in signal loss during bolus passage. In the parameter maps truncation artefacts were significantly reduced using keyhole reconstruction. The keyhole method is appropriate for enhancing image quality in perfusion-weighted imaging on standard imagers without sacrificing time resolution or information about transitory susceptibility changes. However, it should be applied carefully, because the spatial resolution of the dynamic signal change and the cerebrovascular parameters is less than that afforded by the spatial resolution of the reconstructed images. Received: 31 August 2000 Accepted: 5 December 2000     

A 2D MTF approach to evaluate and guide dynamic imaging developments

As the number and complexity of partially sampled dynamic imaging methods continue to increase, reliable strategies to evaluate performance may prove most useful. In the present work, an analytical framework to evaluate given reconstruction methods is presented. A perturbation algorithm allows the proposed evaluation scheme to perform robustly without requiring knowledge about the inner workings of the method being evaluated. A main output of the evaluation process consists of a two‐dimensional modulation transfer function, an easy‐to‐interpret visual rendering of a method's ability to capture all combinations of spatial and temporal frequencies. Approaches to evaluate noise properties and artifact content at all spatial and temporal frequencies are also proposed. One fully sampled phantom and three fully sampled cardiac cine datasets were subsampled (R = 4 and 8) and reconstructed with the different methods tested here. A hybrid method, which combines the main advantageous features observed in our assessments, was proposed and tested in a cardiac cine application, with acceleration factors of 3.5 and 6.3 (skip factors of 4 and 8, respectively). This approach combines features from methods such as k‐t sensitivity encoding, unaliasing by Fourier encoding the overlaps in the temporal dimension‐sensitivity encoding, generalized autocalibrating partially parallel acquisition, sensitivity profiles from an array of coils for encoding and reconstruction in parallel, self, hybrid referencing with unaliasing by Fourier encoding the overlaps in the temporal dimension and generalized autocalibrating partially parallel acquisition, and generalized autocalibrating partially parallel acquisition–enhanced sensitivity maps for sensitivity encoding reconstructions. Magn Reson Med, 2010. © 2009 Wiley‐Liss, Inc.     

Temporally constrained reconstruction of dynamic cardiac perfusion MRI.

Dynamic contrast-enhanced (DCE) MRI is a powerful technique to probe an area of interest in the body. Here a temporally constrained reconstruction (TCR) technique that requires less k-space data over time to obtain good-quality reconstructed images is proposed. This approach can be used to improve the spatial or temporal resolution, or increase the coverage of the object of interest. The method jointly reconstructs the space-time data iteratively with a temporal constraint in order to resolve aliasing. The method was implemented and its feasibility tested on DCE myocardial perfusion data with little or no motion. The results obtained from sparse k-space data using the TCR method were compared with results obtained with a sliding-window (SW) method and from full data using the standard inverse Fourier transform (IFT) reconstruction. Acceleration factors of 5 (R = 5) were achieved without a significant loss in image quality. Mean improvements of 28 +/- 4% in the signal-to-noise ratio (SNR) and 14 +/- 4% in the contrast-to-noise ratio (CNR) were observed in the images reconstructed using the TCR method on sparse data (R = 5) compared to the standard IFT reconstructions from full data for the perfusion datasets. The method has the potential to improve dynamic myocardial perfusion imaging and also to reconstruct other sparse dynamic MR acquisitions.     

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.     

Multiple region MRI.

Traditional Fourier MR imaging (FT MRI) utilizes the Whittaker-Kotel'nikov-Shannon (WKS) sampling theorem. This theorem specifies the spatial frequency components which need to be measured to reconstruct an image with a known field of view (FOV). In this paper, we generalize this result in order to find the optimal k-space sampling for images that vanish except in multiple, possibly non-adjacent regions within the FOV. This provides the basis for "multiple region MRI" (mrMRI), a method of producing such images from a traction of the k-space samples required by the WKS theorem. Image reconstruction does not suffer from noise amplification and can be performed rapidly with fast Fourier transforms, just as in conventional FT MRI. The mrMRI method can also be used to reconstruct images that have low spatial-frequency components throughout the entire FOV and high spatial frequencies (i.e. edges) confined to multiple small regions. The greater efficiency of mrMRI sampling can be parlayed into increased temporal or spatial resolution whenever the imaged objects have signal or "edge" intensity confined to multiple small portions of the FOV. Possible areas of application include MR angiography (MRA), interventional MRI, functional MRI, and spectroscopic MRI. The technique is demonstrated by using it to acquire Gd-enhanced first-pass 3D MRA images of the carotid arteries without the use of bolus-timing techniques.     

Iterative Next-Neighbor Regridding (INNG): improved reconstruction from nonuniformly sampled k-space data using rescaled matrices.

The reconstruction of MR images from nonrectilinearly sampled data is complicated by the fact that the inverse 2D Fourier transform (FT) cannot be performed directly on the acquired k-space data set. k-Space gridding is commonly used because it is an efficient reconstruction method. However, conventional gridding requires optimized density compensation functions (DCFs) to avoid profile distortions. Oftentimes, the calculation of optimized DCFs presents an additional challenge in obtaining an accurately gridded reconstruction. Another type of gridding algorithm, the block uniform resampling (BURS) algorithm, often requires singular value decomposition (SVD) regularization to avoid amplification of data imperfections, and under some conditions it is difficult to adjust the regularization parameters. In this work, new reconstruction algorithms for nonuniformly sampled k-space data are presented. In the newly proposed algorithms, high-quality reconstructed images are obtained from an iterative reconstruction that is performed using matrices scaled to sizes greater than that of the target image matrix. A second version partitions the sampled k-space region into several blocks to avoid limitations that could result from performing multiple 2D-FFTs on large data matrices. The newly proposed algorithms are a simple alternative approach to previously proposed optimized gridding algorithms.