Image Reconstruction: Dynamic Imaging & Phase Mapping
Wednesday 22 April 2009
Room 313BC 16:00-18:00

Moderators:

 Ricardo Otazo and John Pauly
 
16:00 557. Reconstruction Strategies for MRI with Simultaneous Excitation and Acquisition
    Markus Weiger1, Klaas Paul Pruessmann2, Franek Hennel3
1Bruker BioSpin AG, Faellanden, Switzerland; 2Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland; 3Bruker BioSpin MRI GmbH, Ettlingen, Germany
    The concept of simultaneous excitation and acquisition (SEA) introduced with the SWIFT technique enables MRI of samples with very short T2 also under B1 constraints. Here, new reconstruction strategies are reported that address the specific encoding scheme of SEA in a general way, providing an exact and flexible reconstruction procedure. Artefact-free SEA images acquired with 50 kHz bandwidth and 10 µs echo time are presented.
     
16:12 558. k-T PCA: Temporally Constrained k-T BLAST Reconstruction Using Principal Component Analysis
    Henrik Pedersen1, Sebastian Kozerke2
1Functional Imaging Unit, Glostrup Hospital, Glostrup, Denmark; 2Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland
    The k-t BLAST technique has become widespread for achieving faster dynamic MRI. In its basic form k-t BLAST speeds up the data acquisition by undersampling k-t space and resolves the resulting aliasing in the reciprocal x-f space using an adaptive filter derived from low-resolution training images. However, this filtering process tends to increase the reconstruction error or lower the achievable acceleration factor. We show that temporal basis functions calculated by subjecting the training data to principal component analysis (PCA) can be used to constrain the reconstruction such that the temporal resolution is improved. The presented method is called k-t PCA.
     
16:24 559. Technique for Reconstruction Based on Intensity Order (TRIO) Applied as a Second Stage for Dynamic MRI Reconstruction
    Leonardo Ramírez1,2, Claudia Prieto3, Cristian Tejos1,2, Marcelo Guarini1,2, Pablo Irarrazaval1,2
1Departamento de Ingeniería Eléctrica, Pontificia Universidad Católica de Chile, Santiago, Chile; 2Biomedical Imaging Center, Pontificia Universidad Católica de Chile, Santiago, Chile; 3Division of Imaging Sciences, King's College London, London, UK
    Recently, a Technique for Reconstruction based on Intensity Order (TRIO) has been proposed which uses the intensity order of the pixels of an image to reconstruct undersampled data. This work introduces the use of TRIO as a second stage reconstruction to improve the results obtained from traditional undersampled reconstruction algorithms. The effects of incorrect estimation of the intensity order on the results are also discussed, showing that TRIO reconstruction is dependant on the quality of the intensity order information but always achieving improvements in the reconstruction.
     
16:36 560. Accelerating Dynamic MRI Via Spatially Varying Causal Windows
    Uygar Sümbül1, Juan Manuel Santos, John Mark Pauly1
1Electrical Engineering, Stanford University, Stanford, CA, USA
    A causal, pixel-dependent exponential decay window is suggested to improve time series reconstruction. The study is motivated by the observation that many image pixels change slowly over time, while a few pixels experience rapid changes. The window interpretation is realized via a Kalman filter based algorithm. This fast statistical algorithm decreases the temporal blur of the sliding window reconstruction. Moreover, the algorithm handles arbitrary readout trajectories and multiple coils naturally.
     
16:48 561. Motion Adaptive HYPR: An Algorithm for Dynamic Imaging Applications
    Lauren Keith1, Alexei Samsonov1, Orhan Unal1, Dana Peters2, Charles Mistretta1,3, Julia Velikina1
1Medical Physics, University of Wisconsin-Madison, Madison, WI, USA; 2Harvard University, Boston, MA, USA; 3Radiology, University of Wisconsin-Madison, Madison, WI, USA
    HYPR and HYPR LR algorithms provide means for reconstructing images with high SNR and high spatial resolution from undersampled datasets. However, the conventional formation of the composite image makes applications to dynamic imaging difficult. HYPR MA circumvents the problem of blurring due to object motion by computing a composite image in an alternate fashion. Using this technique, dynamic images can be reconstructed with high SNR, high spatial resolution, and high temporal resolution. Applications to cardiac imaging and catheter tracking are shown.
     
17:00 562. Integration of Higher-Order Dynamic Fields Into MR Image Reconstruction
    Bertram Jakob Wilm1, Christoph Barmet1, Matteo Pavan1, Peter Boesiger1, Klaas Paul Pruessmann1
1Institute for Biomedical Engineering, University and ETH Zurich, Zurich, Switzerland
    Recent improvements in magnetic field monitoring opened up the possibility of capturing higher-order dynamic field evolution, providing the necessary information to correct for the image distortions they cause. In the present work a technique for higher order field reconstruction is introduced and demonstrated for diffusion weighted EPI. It has been found that eddy currents of diffusion gradients can induce considerable higher-order field perturbations, causing residual error in conventional first-order reconstruction. Higher-order reconstruction largely eliminated these errors in phantom and in-vivo experiments.
     
17:12 563. Effects of Discrete and Finite Sampling in PatLoc Imaging
    Gerrit Schultz1, Maxim Zaitsev1, Jürgen Hennig1
1Dept. of Diagnostic Radiology, Medical Physics, University Hospital Freiburg, Freiburg, Germany
    Discrete and finite sampling leads to the well known truncation artifacts, which include ringing and possibly aliasing. In PatLoc imaging the gradients are replaced by nonlinear, non-bijective encoding fields. In this case no trivial mapping from frequency space to image space exists. It turns out that the truncation artifacts appear in frequency space in principle in the usual way. The shape of the encoding fields then determines how these artifacts translate into image space. The purpose of this work is to examine these properties and illustrate them with simulation data. One application of this finding is accelerated imaging using PatLoc and SENSE imaging in combination.
     
17:24 564. Low-Resolution Spectral Cost Function for Field Map Estimation
    Kristin L. Granlund1,2, Bruce L. Daniel1, Brian A. Hargreaves1
1Radiology, Stanford University, Stanford, CA, USA; 2Electrical Engineering, Stanford University, Stanford, CA, USA
    Spiral imaging is very sensitive to field inhomogeneity and, therefore, is significantly improved by off-resonance correction. Generating an accurate field map is essential to correcting for B0 field variations. In this work, we use low-resolution spectra to estimate off-resonance, which is then used to perform multi-frequency water/fat separation for spiral breast imaging.
     
17:36 565.

Rapid Fieldmap Estimation for Cardiac Shimming
    Saurabh Shah1, Peter Kellman2, Andreas Greiser3, Peter J. Weale1, Sven Zuehlsdorff1, Renate Jerecic1
1Siemens Medical Solutions USA, Inc., Chicago, IL, USA; 2National Institutes of Health, Bethesda, MD, USA; 3Siemens AG Healthcare Sector, Erlangen, Germany
    Accurate field map estimation is the first step towards an improved cardiac shimming. Field map estimation in the heart is challenging due to the presence of cardiac and respiratory motion, and blood flow effects. In this study, the effects of cardiac and respiratory motion on field map acquisition were investigated using multi-echo GRE sequence. High resolution field maps acquired at different cardiac phases were analyzed to study the effects of cardiac motion. Different field map acquisition schemes were compared to derive a rapid non-ECG triggered method with parallel imaging support, which provides volumetric coverage around the heart in 5-6 seconds.
     
17:48 566. JIGSAW: Joint Inhomogeneity Estimation Via Global Segment Assembly for Water-Fat Separation
    Yi Lu1, Wenmiao Lu2, Brian Andrew Hargreaves3
1Electrical Engineering, Stanford University, Stanford, CA, USA; 2Electrical & Electronic Engr., Nanyang Technological University, Singapore, Singapore; 3Radiology, Stanford University, Stanford, CA, USA
    Key to the success of three-point water-fat separation is reliable estimation of field inhomogeneities, which remains difficult in many clinical applications. The difficulty arises when the spectral field-of-view is not sufficient to accommodate the field inhomogeneities, causing aliasing. This work describes a novel field map estimation technique called JIGSAW, which is based on belief propagation (BP) to produce large segments of pixels with smooth field map values. The field map estimation problem is then reduced to the assembly of a few large segments. In vivo results show that JIGSAW correctly resolves field inhomogeneities in the presence of spectral aliasing.