Claudia Prieto¹, and Tobias Schaeffter^1
1Division of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom

Accurate visualization of the valve leaflet morphology is needed for the assessment of valvular heart disease. Accurate synchronization of the imaging sequence with the opening and closing of the valve is required. We propose the use of the golden ratio radial trajectory to retrospectively select and reconstruct a high-spatial resolution black-blood image at any specific time-point over the valves dynamics. In a first step, the acquired data is used to reconstruct a dynamic sequence with high-temporal resolution. This information is then used to select the right timing for a high-spatial resolution reconstruction from the same data.

Christoph Kolbitsch¹, Claudia Prieto¹, Jouke Smink², and Tobias Schaeffter^1
1Division of Imaging Sciences and Biomedical Engineering, King's College London, London, United Kingdom, ²Philips Healthcare, Best, Netherlands

The major challenges for 3D whole-heart MRI are long acquisition times and respiratory motion. Here a fast and highly efficient 3D whole-heart acquisition scheme is presented. It combines the ideas of Radial Phase Encoding(RPE) allowing for high acceleration factors and Phase Ordering with Automatic Window Selection(PAWS) yielding high navigator efficiency. A new phase encoding scheme is proposed to ensure a short scan time for isotropic high resolution images. Volunteer scans show a strong decrease in scan time for RPE-PAWS compared to a respiratory gated Cartesian scan and indicate a better depiction of small structures such as the coronary arteries.

Holden H Wu^1,2, Dwight G Nishimura², Michael V McConnell^1,2, and Bob S Hu^2,3
1Cardiovascular Medicine, Stanford University, Stanford, CA, United States, ²Electrical Engineering, Stanford University, Stanford, CA, United States, ³Palo Alto Medical Foundation, Palo Alto, CA, United States

Conventional methods for imaging cardiac function seek to suspend or effectively eliminate respiratory motion to avoid image artifacts. However, in many disease states, including pericardial constriction and diastolic dysfunction, it is precisely the changes in cardiac function associated with changes in respiration that can reflect the pathophysiology. In this work, we present a comprehensive free-breathing technique for capturing the five-dimensional state of the heart, including volumetric spatial information, cardiac phase information, and respiratory phase information. This proposed technique collects data using the 3D cones readout trajectory to reduce scan time and provide robustness to motion/flow effects.

Benjamin Zahneisen¹, Thimo Grotz¹, Maxim Zaitsev¹, and Juergen Hennig^1
1University Hospital Freiburg, Freiburg, Germany

MR-encephalography (MREG) has been shown to allow extremely fast and highly sensitive monitoring of functional activation. Recently, it has been shown that the use of a 3D rosette trajectory provides whole brain coverage with an acquisition time of 23ms. However, the rosette trajectory has a very non-uniform k-space sampling density and suffers from off-resonance artifacts. Here, we propose the use of a single shot, variable density, concentric shells trajectory for ultra-fast functional imaging. The method yields very good spatial localization of BOLD-activation. The high sampling rate allows the real time observation of dynamic changes of the BOLD-response (dynamic retinotopic mapping).

Florian Hoffmann¹, Philipp Ehses², Michael Völker², Felix A Breuer², Martin Blaimer², and Peter M Jakob^1,2
1Department of Experimental Physics 5, University of Würzburg, Würzburg, Bayern, Germany, ²Research Center Magnetic Resonance Bavaria (MRB), Würzburg, Germany

One of the first continuously moving table experiments was helical MRI. It uses a radial readout with a linearly increasing projection angle. Helical MRI requires a linear interpolation reconstruction otherwise the table movement provokes artifacts. However, this smears details into neighboring slices. In this work, the angular sampling was modified based on the golden ratio approach. As k-space is covered almost uniformly at any time it is possible to vary the projection number during a sliding window reconstruction or to apply a KWIC-filter. This results in an improved through-slice resolution as demonstrated by phantom and in vivo experiments.

Mitsuharu Miyoshi¹, Naoyoki Takei¹, Ananth J Madhuranthakam², and Hiroyuki Kabasawa^1
1Global Applied Science Laboratory, GE Healthcare Japan, Hino, Tokyo, Japan, ²Global Applied Science Laboratory, GE Healthcare, Boston, MA, United States

3D Fast Spin Echo with variable flip angle uses low flip angle refocus pulses and long echo trains. The effective TE of FSE corresponds to echo train number of k-space center. To make TE flexible, novel view ordering methods were developed in this paper. Phantom and volunteer were scanned with conventional and novel view orderings. Ringing, blur, signal intensity and contrast were measured and compared.

Samantha J Holdsworth¹, Rafael O'Halloran¹, Stefan Skare², and Roland Bammer^1
1Department of Radiology, Stanford University, Palo Alto, CA, United States, ²Clinical Neuroscience, Karolinska Institute, Stockholm, Sweden

The most commonly used SWI acquisition uses a 3D gradient echo (GRE) sequence, however due to the inefficient coverage of k-space per TR, 3D GRE suffers from a long scan time and even subtle motion can considerably hamper the quality of final SWI image. 3D SAP-EPI has been used as a much faster and motion-robust technique than 3D GRE for SWI. Here, we implement Readout-Segmented (RS)-EPI as an alternative. Due to the unidirectional distortions and reduction of blurring compared to 3D SAP-EPI, we show that 3D RS-EPI is a promising technique for the acquisition of fast SWI images.

J. Andrew Derbyshire¹, Haris Saybasili¹, Liheng Guo², Ozan Sayin², Peter Kellman¹, Robert J. Lederman¹, and Daniel A. Herzka^2
1National Heart, Lung and Blood Institute, NIH, Bethesda, MD, United States, ²Department of Biomedical Engineering, Johns Hopkins School of Medicine, Baltimore, MD, United States

Golden Step phase encoding provides close-to-uniform sampling of k-space for any number of consecutive TRs. We demonstrate a MR imaging strategy that allows simultaneous real-time and ECG-gated Cine imaging from a single acquisition. In particular, multiple sets of images with differing temporal resolutions may be reconstructed with arbitrary acceleration rates at arbitrary temporal positions from a single Cartesian acquisition. Furthermore, the gated Cine reconstructions can be performed from acquisitions comprising an arbitrary number of heartbeats, permitting the reconstruction of images even from incomplete breath-holds.

Naoharu Kobayashi¹, Steen Moeller¹, Jang-Yeon Park², and Michael Garwood^1
1Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, United States, ²School of Biomedical Engineering, College of Biomedical and Health Science, Konkuk University, Chungju, Korea, Republic of

A Bloch equation based algebraic reconstruction method applied to concurrent dephasing and excitation (CODE) with frequency-modulated (FM) pulse excitation is introduced. Compensation of the quadratic phase generated by FM excitation has been performed based on a linear approximation of the spin system. However, the actual spin system has non-linearity in its time evolution that is described by the Bloch equation. Therefore, incorporation of the non-linearity into the reconstruction method should improve the image quality. Here, we show the computer simulation and reconstructed images of experimental data acquired on a 16.4 T MRI system.

Masahiro Takizawa¹, Hikaru Hanada¹, Kuniharu Oka¹, and Tetsuhiko Takahashi^1
1MRI system division, Hitachi Medical Corporation, Kashiwa, Chiba, Japan

The UTE sequence, based on radial sampling, acquires echo signal from central to outer parts of k-space. For this kind of sequence, K-trajectory error caused by gradient system response becomes a major problem. For clinical use of UTE it is important to be able to change imaging conditions and parameters for a general oblique imaging plane without causing K-trajectory errors. Even if there are errors on the gradient output, the K-trajectory can be estimated by calculating the error strictly. In this study, the multiple function models are used for correcting the total error of the gradient system response.

Iman Aganj¹, Christophe Lenglet², Essa Yacoub², Guillermo Sapiro¹, and Noam Harel^2
1Electrical Engineering, University of Minnesota, Minneapolis, MN, United States, ²Center for Magnetic Resonance Research, University of Minnesota, United States

Hardware, timing, and SNR considerations restrict the slice-selection and the in-plane resolutions differently, generally resulting in thicker acquisition slices and therefore anisotropic voxels. This non-uniform sampling can be problematic, especially in image segmentation and clinical examination, since the image will be missing high frequencies in the slice-selection direction. High-resolution MR volumes with isotropic voxels are acquired at the cost of requiring the subject to be motionless for a clinically unreasonably long time, hence increasing the risk of motion artifacts. This can be alleviated by dividing the acquisition into (two or) three separate scans, with thicker slices yet complementary resolutions. Every scan will then have a shorter acquisition time and a lower chance of undergoing motion-related distortion. Misalignment between the three scans can be corrected by employing a variety of available registration techniques, and the high-resolution image is eventually reconstructed from the three scans. In this work, we adopt a non-iterative wavelet-based approach, which takes into account the actual system response of the MR scanner. We show results from three orthogonal Susceptibility-Weighted Imaging datasets acquired at 7T, and compare them with a high resolution ground truth dataset.

Keumsil S Lee^1,2, Werner W Roeck^1,3, Grant T Gullberg⁴, and Orhan Nalcioglu^1,3
1Tu & Yuen Center for Functional Onco-Imaging, University of California, Irvine, Irvine, CA, United States, ²Department of Electrical Engineering and Computer Science, University of California, Irvine, Irvine, CA, United States, ³Department of Radiological Sciences, University of California, Irvine, Irvine, CA, United States, ⁴Ernest Orlando Lawrence Berkeley National Laboratory, Berkeley, CA, United States

In nuclear imaging, a limited field of view (LFOV) system is more practical due to the fact that the ROIs are usually a lot smaller than the whole subject and the detectors to build a gamma camera are expensive. However, image reconstruction for the projections acquired by LFOV imaging systems have been done using standard full-field-of-view algorithm. The work presented here proposes a LFOV image reconstruction method named Keyhole SPECT (K-SPECT) that uses anatomical a priori information of ROI determined by high-resolution MR images and radioactivity distribution from SPECT image reconstructed without any a priori information. The simulation results indicate that K-SPECT reconstruction improved the image quality and the accuracy.

Sarah E. Riske¹, Amir Eissa¹, Sandra M. Meyers¹, and Alan H. Wilman^1
1University of Alberta, Edmonton, Alberta, Canada

A means to perform phase filtering in susceptibility images is introduced and applied to MS patients for lesion visualization and for deep grey matter. The method is based on local polynomial fitting using a moving window and provides improved lesion and deep grey matter contrast.

W. Scott Hoge¹, Hong Pan¹, Huan Tan², Emily Stern¹, and Robert A. Kraft^2
1Radiology, Brigham and Women's Hospital, Boston, MA, United States, ²Virginia-Tech Wake Forest School of Biomedical Engineering, Winston-Salem, NC, United States

Echo planar imaging often suffers from signal dropout in regions with high susceptibility. This is particularly problematic for fMRI studies of the brain near the nasal sinuses. Z-shim methods are one approach to recover this lost signal, where a z gradient is employed prior to the EPI readout to counter phase accumulation in the signal dropout region. Single-shot z-shim methods give improved temporal resolution, but with a cost of longer echo trains, which increases geometric distortion. Here, we employ parallel MR imaging (pMRI) to shorten the EPI echo train in a single-shot z-shim method. Temporal encoding complements the approach, to counter Nyquist ghost effects and improve the pMRI calibration. Temporal signal stability is shown to be comparable to a two-shot z-shim method.

Nicholas Ryan Zwart¹, and James Grant Pipe^1
1Neuroimaging Research, Barrow Neurological Institute, Phoenix, Arizona, United States

The reconstruction of non-cartesian k-space trajectories often requires the estimation of non-uniform sampling density. Particularly for 3D, this calculation can be computationally expensive. The method proposed in this work is the combination of a fast, previously proposed, iterative algorithm with a previously proposed optimized convolution kernel.

W. Scott Hoge¹, Huan Tan², Zhikui Xiao³, and Robert A. Kraft^2
1Radiology, Brigham and Women's Hospital, Boston, MA, United States, ²Virginia-Tech Wake Forest School of Biomedical Engineering, Winston-Salem, NC, United States,³Global Applied Science Laboratory, GE Healthcare, Beijing, China, People's Republic of

GESTE is a robust method of ghost correction that employs temporal encoding with parallel imaging. In this work we demonstrate that the temporal encoding requirements can be applied across the slice dimension in a single volume acquisition. Working from the assumption that the signal changes slowly across slices, this enables accurate GRAPPA parameters to be estimated and used in a GESTE-style image reconstruction pipeline. The method is shown to be effective in in-vivo brain EPI images, with only minimal errors introduced compared to a standard GESTE reconstruction.

Nicole Seiberlich¹, Gregory Lee¹, Philipp Ehses², Jeffrey Duerk^1,3, and Mark Griswold^1,3
1Radiology, University Hospitals of Cleveland, Cleveland, OH, United States, ²Research Center for Magnetic Resonance Bavaria (MRB), Wuerzburg, Germany, ³Biomedical Engineering, Case Western Reserve University, Cleveland, OH, United States

Through-time non-Cartesian GRAPPA has recently emerged as a technique for real-time MRI, where images acquired using an undersampled radial trajectory could be reconstructed to yield a temporal resolution better than 50 ms/image. This through-time GRAPPA method has been extended to the more efficient variable density spiral trajectory in order to scan at a higher frame rate with a lower acceleration factor than with radial. Using through-time spiral GRAPPA and highly undersampled spiral data (R=12), short axis cardiac SSFP images can be generated with a temporal resolution of 17.9ms/image (a frame rate of 56 images/s) without any echo sharing.

Steen Moeller¹, Curtis A Corum¹, Djaudat Idiyatullin¹, and Michael Garwood^1
1University of Minnesota, Minneapolis, 55455, United States

Parallel imaging has been combined with the 3D radial SWIFT sequence and applied to reduced FOV imaging for increased spatial resolution. With a 2 channel coil, an R=2 reconstruction has been demonstrated.

Yuchou Chang¹, Dong Liang¹, and Leslie Ying^1
1Electrical Engineering, University of Wisconsin - Milwaukee, Milwaukee, Wisconsin, United States

We improve the convolution model in GRAPPA using a kernel approach. We map the acquired k-space data through a nonlinear transform to a high-dimensional space and then linearly combine the transformed data to estimate the missing k-space data. Both polynomial and Gaussian kernels are investigated for the nonlinear transform. The proposed kernel model characterizes the system noise in reconstruction more accurately. Experimental results using in vivo data demonstrate that the proposed kernel GRAPPA method can significantly suppress the noise in conventional GRAPPA without introducing artifacts.

Haifeng Wang¹, Dong Liang¹, Kevin F. King², Gajanan Nagarsekar¹, and Leslie Ying^1
1Department of Electrical Engineering and Computer Science, University of Wisconsin-Milwaukee, Milwaukee, WI, United States, ²Global Applied Science Lab, General Electric Healthcare, Waukesha, WI, United States

A cross sampling method is proposed to acquire the ACS lines orthogonal to the reduced lines for GRAPPA. This cross sampling method increases the amount of calibration data along the direction that the k-space is undersampled and thus improves the calibration accuracy, especially when a small number of ACS lines are acquired. Both phantom and in vivo experiments demonstrate that the proposed method, named cross-sampled GRAPPA (CS-GRAPPA), can effectively reduce the aliasing artifacts of GRAPPA when high acceleration is desired.

Xiao-Long Wu¹, Jiading Gai², Fan Lam^1,2, Maojing Fu^1,2, Justin P. Haldar^1,2, Yue Zhuo^2,3, Zhi-Pei Liang^1,2, Wen-Mei Hwu^1,2, and Bradley P. Sutton^2,3
1Electrical and Computer Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ²Beckman Institute, University of Illinois at Urbana-Champaign, Urbana, IL, United States, ³Bioengineering Department, University of Illinois at Urbana-Champaign, Urbana, IL, United States

Despite advances in acquisition and reconstruction technologies, typical clinical scans rely on Cartesian acquisitions and limited reconstruction routines. Requirements for significant computational resources and specialized expertise are a barrier to widespread use of algorithms that combine efficient non-Cartesian trajectories, field inhomogeneity correction, parallel imaging, and image regularization. We present a parallel implementation of such a reconstruction utilizing manycore graphics processing cards to speed reconstruction to acceptable levels, even for large matrix sizes and multiple coil acquisitions. We compare reconstruction times with parallel C-code and a common approximation method, showing that the proposed code is faster without using interpolation operators.

Yu Li¹, and Charles L. Dumoulin^1
1Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio, United States

This work introduces a novel parallel imaging technique that uses a non-uniform k-space undersampling trajectory to accelerate Cartesian data acquisition for MRI. In this technique, a non-uniform undersampling trajectory is formed from multiple uniform sub-trajectories with higher undersampling factors. A modulation model is used to separate non-uniformly undersampled data into aliasing and real image data in k-space. A set of linear filters is used to pass real image and block aliasing for reconstruction. In high-resolution brain imaging, it is demonstrated that a non-uniform trajectory gives better performance in parallel imaging with high acceleration factors than a uniform trajectory.

Jacob R. Hoberg¹, and Nan-Kuei Chen^1
1BIAC, Duke University, Durham, NC, United States

The use of the multi-band parallel imaging along the slice-select direction in fMRI can increase the throughput without signal reduction resulting from k-space undersampling as in conventional parallel acceleration along the phase-encoding direction. However, the multi-band parallel imaging is still susceptible to noise amplification due to non-ideal sensitivity profile. The existing methods may not improve the SNR when the simultaneously excited bands are not spatially far apart. The purpose of this work is to present a novel method that integrates Hadamard slice encoding and time-domain phase constrained reconstruction, to eliminate the noise amplification of multi-band parallel.

Zarko Celicanin¹, Frank Preiswerk², Patrik Arnold², Philippe Cattin², Klaus Scheffler¹, and Francesco Santini^1
1Radiological Physics, University of Basel Hospital, Basel, Switzerland, ²Medical Imaging Analysis Center, University of Basel, Basel, Switzerland

Respiratory organ motion is a complicating factor in many treatments. Novel respiratory correlated MR imaging method was recently purposed that acquires a full 2D sagittal navigator image and builds 4D-MRI organ model. Image and navigator slices were acquired interleaved. Since the navigator image is used to estimate the exact organ position and shape, a time lag could lead to discrepancy in images and to incorrect models. We describe a new approach to image and navigator acquisition based on multislice CAIPIRINHA, which allows a true simultaneous acquisition with consequent decrease of lag and increase of temporal resolution.

Yu Ding¹, Yiu-Cho Chung², and Orlando Simontetti^3
1The Ohio State University, Columbus, OH, United States, ²Siemens Medical Solutions, ³The Ohio State University

Image based parallel magnetic resonance imaging (pMRI) techniques (SENSE or its variants) use the best linear unbiased estimation (BLUE) to reconstruct image. Mathematically, the signal-to-noise-ratio (SNR) of images reconstructed by BLUE is better or equal to the SNR of the simple least square solution, depending on the accuracy of the noise covariance matrix used. Hence, the accuracy of the noise covariance matrix estimation affects the SNR performance of pMRI. In this study, we use volunteer study to quantify the how the errors of covariance matrix estimation affect image SNR, and propose a guideline for accurate covariance matrix estimation.

Fa-Hsuan Lin^1,2, Panu Vesanen³, Jaakko O. Nieminen³, John W. Belliveau², and Risto J. Ilmoniemi^3
1Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan, ²Martinos Center, Massachusetts General Hospital, Charlestown, MA, United States,³Department of Biomedical Engineering and Computational Science (BECS), Aalto University, Espoo, Finland

In ultra-low-field (ULF) magnetic resonance imaging (MRI), a readily available sensory array consisting of up to hundreds of detectors can be used for simultaneous data acquisition. Such a parallel data acquisition immediately allows the application of the parallel MRI principle to reduce the data acquisition time, which critically depends on the number of magnetization polarization steps in ULF-MRI. In this study, we investigate the theoretical signal-to-noise ratio (SNR) penalty based on the Sensitivity Encoded (SENSE) MRI at various acceleration rates using the optimized MEG array with different channels of pick-up coils.

Hammad Omer¹, and Robert Dickinson^2
1Imperial College London, London, London, United Kingdom, ²Imperial College London

Parallel MRI is a technique to reduce the scan time for MRI image acquisition but this comes at the cost of noise amplification during the process of image reconstruction. A method based on the use of g-factor as a regularization parameter in Tikhonov regularized SENSE reconstruction is proposed. The results show a considerable improvement in the reconstructed image as compared to contemporary methods. It has been shown that the g-factor map effectively acts as a spatially adaptive regularization parameter providing very good reconstruction at much less computational cost.

Lei Hou Hamilton¹, Sumati Krishnan², Senthil Ramamurthy², David Moratal³, and Marijn Brummer^2
1School of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, United States, ²School of Medicine, Emory University, Atlanta, GA, United States, ³Center for Biomaterials and Tissue Engineering, Universitat Politècnica de València, Valèncian, Spain

A simplified variant of the PINOT method is presented, which combines parallel imaging and iterative Noquist (iNoquist) sequentially to greatly accelerate the reconstruction speed. Compared with parallel imaging alone with the same reduction factor, this approach reduces noise and residual artifacts. Compared with original PINOT with the same reduction factor, it has the advantage of alleviating computational burden, albeit at some penalty in image quality. Furthermore, although illustrated here only for SPACE-RIP, we note that iNoquist can be used as a post processing method following any parallel imaging methods to suppress residual noise and artifacts.

Hanno Homann¹, Tim Nielsen², Kay Nehrke², Ingmar Graesslin², Olaf Dössel¹, and Peter Börnert^2
1Institute of Biomedical Engineering, Karlsruhe Institute of Technology, Karlsruhe, Germany, ²Philips Research Europe, Hamburg, Germany

Most transmit (TX) arrays used in parallel transmission can also be used as receive (RX) arrays, raising the question of optimal signal combination to recover RX field homogeneity. In this study, different approaches to signal combination are compared, using an 8-channel TX/RX body coil at 3T. The possibility to combine the signals of the individual channels in the image space resulted in greatly improved homogeneity compared to a single channel birdcage coil.

Irene Paola Ponce Garcia¹, Martin Blaimer², Felix Breuer², Peter M Jakob^1,2, Mark A Griswold³, and Peter Kellman^4
1Experimental Physics 5, University of Würzburg, Würzburg, Bavaria, Germany, ²Research Center Magnetic Resonance Bavaria e.V (MRB), Würzburg, Bavaria, Germany,³Department of Radiology, University Hospitals of Cleveland and Case Western Reserve University, Cleveland, Ohio, United States, ⁴Laboratory of Cardiac Energetics, National Institutes of Health, National Heart, Lung and Blood Institute, Bethesda, Maryland, United States

In Auto-calibrated Dynamic Parallel Magnetic Resonance Imaging (pMRI), such TSENSE or kt-SENSE, the missing information is reconstructed using the spatial sensitivities of multiple receiver coils. The coil sensitivities are derived from the full FOV temporal average image (a.k.a direct current, DC) that is obtained by averaging the undersampled time frames. In Cartesian sampling, the averaging leads to aliasing artifacts and therefore to errors in the coil sensitivities estimates. As a result, the reconstructed images exhibit temporal filtering effects. In this work, we demonstrate that these temporal filtering effects are not significantly present in accelerated dynamic parallel MRI experiments using Radial sampling.

James H Holmes¹, Kang Wang², Philip J Beatty³, Reed F Busse⁴, Frank R Korosec⁵, Lauren A Keith², Christopher J Francois⁶, Scott B Reeder⁵, and Jean H Brittain^7
1Global Applied Science Laboratory, GE Healthcare, Madison, WI, United States, ²Medical Physics, University of Wisconsin-Madison, Madison, WI, ³Global Applied Science Laboratory, GE Healthcare, Thornhill, ON, Canada, ⁴MR Research, GE Healthcare, Waukesha, WI, ⁵Radiology, University of Wisconsin-Madison, Madison, WI, ⁶Radiology, University of Wisconsin-Madison, Madision, WI, ⁷Global Applied Science Laboratory, GE Healthcare, Madison, WI

An interleaved variable density (IVD) Cartesian acquisition combined with a constrained reconstruction (HYCR) and data-driven parallel imaging has been demonstrated for improved MRA temporal resolution. There are several potential methods for calibrating the parallel imaging including using mask data, mask data subtracted from the time-frame, and data from multiple time-frames. In this work, we present results from these different calibration methods and demonstrate the advantages of using multiple calibration data sources to address data inconsistencies and maintain high temporal resolution with time-resolved MRA.

Suhyung Park¹, Jin-Suck Suh^1,2, and Jaeseok Park^2
1Medical Science, Yonsei University, Seoul, Korea, Republic of, ²Radiology, Yonsei University, Seoul, Korea, Republic of

It is challenging to estimate accurate convolution kernel in k-space based parallel imaging. This work suggest adaptive self-calibrating method designing static calibration region into dynamic calibration region using the Kalman filter