Jinhua Sheng

Joint image reconstruction and sensitivity estimation in SENSE (JSENSE).

Parallel magnetic resonance imaging (pMRI) using multichannel receiver coils has emerged as an effective tool to reduce imaging time in various applications. However, the issue of accurate estimation of coil sensitivities has not been fully addressed, which limits the level of speed enhancement achievable with the technology. The self‐calibrating (SC) technique for sensitivity extraction has been well accepted, especially for dynamic imaging, and complements the common calibration technique that uses a separate scan. However, the existing method to extract the sensitivity information from the SC data is not accurate enough when the number of data is small, and thus erroneous sensitivities affect the reconstruction quality when they are directly applied to the reconstruction equation. This paper considers this problem of error propagation in the sequential procedure of sensitivity estimation followed by image reconstruction in existing methods, such as sensitivity encoding (SENSE) and simultaneous acquisition of spatial harmonics (SMASH), and reformulates the image reconstruction problem as a joint estimation of the coil sensitivities and the desired image, which is solved by an iterative optimization algorithm. The proposed method was tested on various data sets. The results from a set of in vivo data are shown to demonstrate the effectiveness of the proposed method, especially when a rather large net acceleration factor is used. Magn Reson Med 57:1196–1202, 2007.

Regularized sensitivity encoding (SENSE) reconstruction using bregman iterations

In parallel imaging, the signal‐to‐noise ratio (SNR) of sensitivity encoding (SENSE) reconstruction is usually degraded by the ill‐conditioning problem, which becomes especially serious at large acceleration factors. Existing regularization methods have been shown to alleviate the problem. However, they usually suffer from image artifacts at high acceleration factors due to the large data inconsistency resulting from heavy regularization. In this paper, we propose Bregman iteration for SENSE regularization. Unlike the existing regularization methods where the regularization function is fixed, the method adaptively updates the regularization function using the Bregman distance at different iterations, such that the iteration gradually removes the aliasing artifacts and recovers fine structures before the noise finally comes back. With a discrepancy principle as the stopping criterion, our results demonstrate that the reconstructed image using Bregman iteration preserves both sharp edges lost in Tikhonov regularization and fines structures missed in total variation (TV) regularization, while reducing more noise and aliasing artifacts. Magn Reson Med 61:145–152, 2009.

Metabolomic imaging for human prostate cancer detection.

Magnetic resonance spectroscopy can generate comprehensive maps of metabolites in the prostate that reveal cancerous regions, potentially guiding biopsies and prognosis. Imaging Prostate Cancer: Now You See It Getting regular checkups for prostate cancer can give a sense of security to men, but the tests used for this screening are not infallible. High concentrations of PSA (prostate-specific antigen) in the blood are a red flag but can be caused by several benign conditions. Even the gold standard, a prostate biopsy, only reveals whether cancer is present in the 8 to 16 small regions of the gland into which the tiny biopsy needle was inserted; tumors between the biopsy sites are often missed. As reported here by Wu et al., sophisticated imaging methods may soon help to improve this diagnostic dilemma by creating three-dimensional metabolic maps of the prostate that pinpoint cancerous regions. Magnetic resonance spectroscopy can examine an array of compounds within living cells by pulsing the intact tissue with radio-frequency pulses in a magnetic field and measuring the collective reactions of the resident metabolites. A normal prostate will yield a characteristic complex spectrum of metabolites, whereas cancerous tissue will yield an equally complex but altered spectrum. The authors used principal components analysis and discriminant analysis—statistical methods used to extract information from complex data— to identify portions of the signal that could best distinguish normal tissue from tumor tissue and, from this analysis, they calculated a malignancy index from the spectra. Five surgically removed human prostates, each with confirmed prostate cancer, were imaged in this way with a 7-tesla MRI scanner designed for use on patients. Maps of the malignancy index across the width of the five organs showed a total of 13 hot spots, areas with a high chance of being cancerous. The authors then overlaid a second set of prostate maps identifying tumors in the old-fashioned histological way, by careful staining and study by a pathologist. These histological maps revealed five tumors within the area of the metabolic index maps, each of which corresponded to a region with a high malignancy index by magnetic resonance spectroscopy. The remaining regions with high malignancy indices corresponded to areas that appeared histologically benign. This is an accuracy of 93% when determined on a region-by-region basis. Before clinical radiologists can apply such a technique to human patients, researchers must solve technical challenges associated with imaging the prostate in living patients in high-field 7-tesla magnets. But eventually, the application of metabolite-based imaging could help with detection and accurate biopsy of prostate cancer, and may ultimately provide specific metabolic information to noninvasively inform diagnosis of cancer subtypes and prognosis of the course of the disease. As current radiological approaches cannot accurately localize prostate cancer in vivo, biopsies are conducted at random within prostates for patients at risk for prostate cancer, leading to high false-negative rates. Metabolomic imaging can map cancer-specific biomolecular profile values onto anatomical structures to direct biopsy. In this preliminary study, we evaluated five whole prostates removed during prostatectomy from biopsy-proven cancer patients on a 7-tesla human whole-body magnetic resonance scanner. Localized, multi–cross-sectional, multivoxel magnetic resonance spectra were used to construct a malignancy index based on prostate cancer metabolomic profiles obtained from previous intact tissue analyses with a 14-tesla spectrometer. This calculated malignancy index is linearly correlated with lesion size and demonstrates a 93 to 97% overall accuracy for detecting the presence of prostate cancer lesions, suggesting the potential clinical utility of this approach.

REGULARIZED SENSE RECONSTRUCTION USING ITERATIVELY REFINED TOTAL VARIATION METHOD

SENSE has been widely accepted as one of the standard reconstruction algorithms for parallel MRI. When large acceleration factors are employed, the SENSE reconstruction becomes very ill-conditioned. For Cartesian SENSE, Tikhonov regularization has been commonly used. However, the Tikhonov regularized image usually tends to be overly smooth, and a high-quality regularization image is desirable to alleviate this problem but is not available. In this paper, we propose a new SENSE regularization technique that is based on total variation with iterated refinement using Bregman iteration. It penalizes highly oscillatory noise but allows sharp edges in reconstruction without the need for prior information. In addition, the Bregman iteration refines the image details iteratively. The method is shown to be able to significantly reduce the artifacts in SENSE reconstruction

JOINT ESTIMATION OF IMAGE AND COIL SENSITIVITIES IN PARALLEL SPIRAL MRI

Parallel magnetic resonance imaging (MRI) using multichannel receiver coils has emerged as an effective tool to reduce imaging time in various dynamic imaging applications. However, there are still a number of image reconstruction issues that have not been fully addressed, thereby limiting the level of speed enhancement achievable with the technology. This paper considers the inaccuracy of coil sensitivities in conventional reconstruction methods such as SENSE, and reformulates the image reconstruction problem as a joint estimation of the coil sensitivities and the desired image, which is solved by an iterative algorithm. Experimental results demonstrate the effectiveness of the proposed method especially when large acceleration factors are used

Improved spiral sense reconstruction using a multiscale wavelet model

SENSE has been widely accepted and extensively studied in the community of parallel MRI. Although many regularization approaches have been developed to address the ill-conditioning problem for Cartesian SENSE, fewer efforts have been made to address this problem when the sampling trajectory is non-Cartesian. For non-Cartesian SENSE using the iterative conjugate gradient method, ill- conditioning can degrade not only the signal-to-noise ratio, but also the convergence behavior. This paper proposes a regularization technique for non-Cartesian SENSE using a multiscale wavelet model. The technique models the desired image as a random field whose wavelet transform coefficients obey a generalized Gaussian distribution. The effectiveness of the proposed method has been validated by in vivo experiments.

Improved self-calibrated spiral parallel imaging using JSENSE

Spiral MRI has several advantages over Cartesian MRI such as faster acquisitions and reduced demand in gradient. In parallel imaging, spiral trajectories are especially of great interest due to their inherent self-calibration capabilities, which is especially useful for dynamic imaging applications such as fMRI and cardiac imaging. The existing self-calibration techniques use the central spiral data that are sampled densely in the accelerated acquisition for coil sensitivity estimation. However, the resulting sensitivities are not sufficiently accurate for SENSE reconstruction due to the data truncation. In this paper, JSENSE which has been successfully used in Cartesian trajectories is extended to spiral trajectory such that the coil sensitivities and the desired image are reconstructed jointly to improve accuracy through alternating optimization. The improved sensitivities lead to a more accurate SENSE reconstruction. The results from both phantom and in vivo data are shown to demonstrate the effectiveness of JSENSE for spiral trajectory.

A variable projection approach to parallel magnetic resonance imaging

Accurate estimation of channel sensitivity functions is still a challenging problem in parallel imaging. JSENSE has recently been proposed to improve the accuracy of sensitivity estimation using the self calibration data. It regards both the coil sensitivities and the desired images as unknowns to be solved for jointly. The existing algorithm for the underlying nonlinear optimization problem requires an accurate initial value, which needs considerable number of self calibration data. In this paper, we use the variable projection method to find the optimal solution. The method is known to be able to give an optimal solution, and our implementation has a linear convergence rate. The performance of the proposed method is evaluated using a set of in vivo experiment data.

Joint Estimation of Coil Sensitivities and Image in Parallel Magnetic Resonance Imaging

Parallel magnetic resonance imaging using multi-channel receiver coils has emerged as an effective tool to reduce imaging time in various applications. In general, exact knowledge of the coil sensitivities is needed to reconstruct the desired image. However, the issue of accurate estimation of coil sensitivities has not been fully addressed; thereby the errors limit the image quality achievable with the technology. In this paper, we model the image reconstruction problem as a blind multi-channel image restoration problem such that both the coil sensitivities and the desired images are estimated jointly. Specifically, a parametric model is assumed for the coil sensitivities whose parameters are solved by an iterative optimization algorithm along with image reconstruction. The proposed method has been tested on various data sets. The results from some in vivo data sets are shown to demonstrate the effectiveness of the proposed method especially when a rather large net acceleration factor is used.