Image Analysis
Tuesday 21 April 2009
Room 312 10:30-12:30


Qi Duan and Simon K. Warfield

10:30  260. MR-Based Attenuation Correction for PET/MR
    Matthias Hofmann1,2, Florian Steinke1, Ilja Bezrukov1,2, Armin Kolb2, Philip Aschoff2, Matthias Lichy2, Michael Erb2, Thomas Nägele2, Michael Brady3, Bernhard Schölkopf1, Bernd Pichler2
Max Planck Institute for Biological Cybernetics, Tuebingen, Germany; 2Department of Radiology, University of Tuebingen, Tuebingen, Germany; 3Wolfson Medical Vision Laboratory, University of Oxford, Oxford, UK
    There has recently been a growing interest in combining PET and MR. Attenuation correction (AC), which accounts for radiation attenuation properties of the tissue, is mandatory for quantitative PET. In the case of PET/MR the attenuation map needs to be determined from the MR image. This is intrinsically difficult as MR intensities are not related to the electron density information of the attenuation map. Using ultra-short echo (UTE) acquisition, atlas registration and machine learning, we present methods that allow prediction of the attenuation map based on the MR image both for brain and whole body imaging.
10:42 261. High Resolution Phase Gradient Mapping as a Tool for the Detection and Analysis of Local Field Disturbances
    Hendrik de Leeuw1, Peter Roland Seevinck1, Gerrit Hendrik van de Maat1, Chris J.G. Bakker1
Image Sciences Institute, University Medical Center Utrecht, Utrecht, Netherlands
    We have demonstrated a post processing technique -phase gradient mapping- that enables us to generate positive contrast of magnetically labeled substances. It furthermore can be used as a tool for the detection and analysis of local field inhomogeneities. The technique can be applied without loss of resolution, does not need phase unwrapping and allows simple error estimations. Phase gradient mapping provides quantitative values and may be expected to find application in many areas, e.g., in studies concerned with the quantification or characterization of field distortions due to contrast agents


10:54 262. Limit of Detection of Localized Absolute Changes in CBF Using Arterial Spin Labeling (ASL) MRI
    Iris Asllani1, Ajna Borogovac, Truman R. Brown, Joy Hirsch, John W. Krakauer
1PICS, Radiology, Columbia University, New York, NY , USA
    In Arterial Spin Labeling (ASL) MRI, the partial volume effects (PVE) are exacerbated by the nonlinear dependency of the ASL signal on voxel tissue heterogeneity. We have developed a method that corrects for PVE in ASL. The method is based on a model that represents the voxel intensity as a weighted sum of pure tissue contribution where the weighting coefficients are the tissue’s fractional volume in the voxel. Here we show the feasibility of this method to quantify absolute changes in CBF. Results from data simulation as well as experimental model from stroke are presented.
11:06 263. An Analytical Model of Diffusion and Exchange of Water in White Matter from Diffusion-MRI and Its Application in Measuring Axon Radii
    Wenjin Zhou1, David H.  Laidlaw1
Computer Science, Brown University, Providence, RI, USA
    We present an analytical model of diffusion and water exchange in white matter to estimate axon radii. Direct measurement of important biomarkers such as the axon radii, density, and permeability are important for early detection of diseases. We use a model with two compartments between which there is exchange of water molecules. Our analytical formulas examine the derivation of axonal parameters that affect the signal attenuation of diffusion-MRI experiments. The model is fitted to Monte Carlo simulation data. The parameters recovered are compared with ground truth from simulation and prove the feasibility of recovering underlying axonal radii using the model.
11:18 264. SMRI Complex Framework for Evaluating Relative Gray and White Matter Group Differences
    Lai Xu1,2, Vince D. Calhoun1,2
The MIND Research Network, Albuquerque, NM, USA; 2Electrical and Computer Engineering, The University of New Mexico, Albuquerque, NM, USA
    In this study, we built a framework for jointly processing gray and white matter data which incorporates a complex-valued construction into source based morphometry (SBM) to identify the sources revealing relative gray and white matter group difference. The framework was applied to a large dataset from schizophrenia patients and healthy controls. Interestingly, some source patterns looked similar to functional MRI patterns which suggested structural brain information underlying functional areas might be identified. Our approach provides a way to jointly identify changes in both gray and white matter and may prove to be a useful tool to study the brain.
11:30 265. Sensitive and Noise-Resistant Identification of Voxel-Wise Changes in Magnetization Transfer Ratio Via Cluster Enhancement and M-Estimator-Based Monte Carlo Simulation
    Michael G. Dwyer1, Niels Bergsland1, Sara Hussein1, Jacqueline Durfee1, Bianca Weinstock-Guttman1,2, Robert Zivadinov1,2
Buffalo Neuroimaging Analysis Center, State University of New York, Buffalo, NY, USA; 2The Jacobs Neurological Institute, State University of New York, Buffalo, NY, USA
    We propose a sensitive and reliable means for automatically quantifying the volumes of significantly increasing and decreasing MTR in brain MRI over time. This method takes advantage of the newly developed threshold-free cluster enhancement (TFCE) technique to increase sensitivity without sacrificing specificity. It also utilizes a Monte Carlo simulation approach to ensure that results can be interpreted within a correct statistical framework. The method is validated via comparative analysis of a patient group with multiple sclerosis and a group of healthy volunteers


11:42 266. A Method for Correcting Inter-Series Motion in Brain MRI for Auto Scan Plane Planning
    Xiaodong Tao1, Sandeep Narendra Gupta1
Visualization and Computer Vision Lab, GE Global Research Center, Niskayuna, NY, USA
    Patient motion is a major problem in MR exams. We propose here an algorithm that relies on the known position and orientation of the anatomy at the beginning of a scan and uses fast three-plane localizers to update this just before image acquisition. The algorithm finds a rigid transform that best aligns the 3P localizers to the full initial volumetric localizer. This transform is then used to compute new patient-centric scan plane prescription. We have incorporated this approach in a clinical MR system and demonstrated its usefulness in automatically obtaining consistent imaging planes in brain exams in presence of motion.
11:54 267. Development and Evaluation of a Quantitative Brain Atlas @1.5T and Its Application to MS

Veronika Ermer1, Heiko Neeb2, N. Jon Shah1,3
Institute of Neurosciences and Biophysics, Research Centre Juelich, Juelich, Germany; 2RheinAhrCampus Remagen, University of Applied Sciences Koblenz, Remagen, Germany; 3Faculty of Medicine, Department of Neurology, RWTH Aachen University, JARA, Aachen, Germany

    The use of standard brain atlases is well established in the MR community, but none of the commonly utilised standard brains or atlases provide quantitative information. Especially for human brain imaging, qMRI is an attractive method to study changes in the brain caused by diseases. Within the framework of qMRI, this work reports on the development of the first quantitative brain atlas for tissue water content. This atlas can be used as a reference for the comparison of the absolute water content of the brain of patients with pathological changes to that in healthy volunteers.


12:06 268. Brain Tissue Segmentation Using Fast T1 Mapping

Wanyong Shin1, Geng Xiujuan1, Hong Gu1, Yihong Yang1
Neuroimaging Research Branch, National Institute on Drug Abuse, Baltimore, MD, USA

    In this study, an automated brain tissue segmentation method based on modeling of individual quantitative T1 values of brain tissues is proposed. To accomplish it, a fast T1 mapping using inversion recovery Look-Locker echo-planar imaging at a steady state (IR LL-EPI SS) with whole brain coverage is presented. This method is insensitive to instrumental settings and can be used to address specific patient populations and age-dependent groups.
12:18 269. Efficient Anatomical Labeling by Statistical Recombination of Partially Label Datasets
    Bennett Allan Landman1, John Anton Bogovic2, Jerry Ladd Prince2,3
Biomedical Engineering, Johns Hopkins University, Baltimore, MD, USA; 2Electrical and Computer Engineering, Johns Hopkins University, Baltimore, MD, USA; 3Biomedical Engineering, Johns Hopkins University, Baltimore, MD, USA
    Manual labeling of medical imaging data is critical task for the assessment of volumetric and morphometric changes; however, even expert raters are imperfect and subject to variability. Existing techniques to combine data from multiple raters require that each rater generate a complete dataset. We propose a robust extension which allows for missing data, accounts for repeated tasks, and utilizes training data. With our technique, numerous raters can label small, overlapping portions of a large dataset, and rater heterogeneity can be robustly controlled while estimating a single, reliable label set. This enables “parallel processing” and reduces detrimental impacts of rater unavailability.