Advances in Image Analysis
Click on to view the abstract pdf and click on to view the video presentation.
Thursday May 12th
Room 511D-F  10:30 - 12:30 Moderators: Shannon Agner and Benoit Scherrer

10:30 531.   Quantitative MRI biomarkers for Knee Pain and Other Symptoms 
Jose Tamez-Pena1, Patricia Gonzalez2, Joshua Farber2, Edward Schreyer2, Saara Totterman2, and Victor Trevino1
1Biomedicine, ITESM, Monterrey, Nuevo Leon, Mexico, 2Qmetrics Technologies, Rochester, NY, United States

This work describes the performance of a fully automated MRI image analysis technique to detect or predict clinical symptoms of knee Osteoarthritis.

10:42 532.   Measuring the volumes and thickness of hippocampal subfields in vivo using automatic segmentation of T2-weighted MRI: A pilot evaluation study 
Paul A. Yushkevich1, Hongzhi Wang1, John Pluta1, Sandhitsu R Das1, Brian Avants1, Michael Weiner2, Susanne Mueller2, and David Wolk3
1PICSL, Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States, 2Department of Radiology, University of California, San Francisco, San Francisco, CA, United States, 3Department of Neurology, University of Pennsylvania, Philadelphia, PA, United States

High-resolution T2-weighted MRI has been shown to be a promising modality for imaging the subfields of the hippocampal formation. However, until now, the analysis of this data required tedious manual segmentation. We present a reliable method for automatic segmentation of hippocampal subfields in T2-weighted MRI and apply it to measure group differences in subfield volume and thickness between amnestic MCI patients and controls.

10:54 533.   MTR at 3T in the Hippocampus – Validation of Automated Post-Analysis and Comparison of Quantification Metrics 
Shawn Sidharthan1, Ryan Joseph Hutten1, Christopher Glielmi2, Hongyan Du3, Fiona Malone1, Ann Ragin1,4, Robert R Edelman1, and Ying Wu1,5
1Radiology, NorthShore University HealthSystem, Evanston, IL, United States, 2MR Research and Development, Siemens Healthcare, Chicago, IL, United States, 3Center for Clinical Research Informatics, NorthShore University HealthSystem, Evanston, IL, United States, 4Radiology, Northwestern University, Chicago, IL, United States,5Radiology, University of Chicago, Chicago, IL, United States

Magnetization transfer ratio (MTR) may detect subtle microscopic changes in the hippocampus before macroscopic anomalies occur. This modality proves to be a crucial tool when evaluating patients with progressive neurological pathologies such as Alzheimer’s disease. In this study, reliability and reproducibility were analyzed for high-resolution MTR at 3T in different metrics (mean and histogram approach) and compared to the more conventional MR volumetric method in the hippocampus. Mean and histogram MTR approach derived from automated post-processing methods, provided excellent scan-rescan results in comparison to volumetry. The results indicate MTR and volumetric analysis to be a useful tool for future studies.

11:06 534.   Analysis of Hippocampal Shape in Children Using a Surface-to-Centerline Distance Method and Template-Based Surface and Volumetric Non-Rigid Registration Methods 
Muqing Lin1, Kevin Head2, Claudia Buss2, Tugan Muftuler1, Elysia Poggi Davis1, Curt A Sandman2, Orhan Nalcioglu1, and Min-Ying Lydia Su1
1Tu & Yuen Center for Functional Onco-Imaging and Department of Radiological Sciences, University of California, Irvine, CA, United States, 2Department of Psychiatry & Human Behavior, University of California, Irvine, CA, United States

The shape analysis of hippocampus was applied to evaluate the changes in developmental brain in 48 children from 6 to 9 years old. Three different methods are used, including the surface-to-centerline distance mapping and non-rigid registration using robust point mapping (RPM) and Demons algorithm. The differences between males and females were also analyzed. Although a significant difference was found in some scattered regions, the averaged difference between the two age or sex groups is very small. The results obtained using the distance mapping and RPM registration methods showed similar patterns. The observed differences are mostly likely coming from individual variations.

11:18 535.   Comparison of tissue classification models for automatic brain MR segmentation 
Delphine Ribes1,2, Bénédicte Mortamet1, Meritxell Cuadra Bach3, Clifford R. Jack4, Reto Meuli5, Gunnar Krueger1, and Alexis Roche1
1Advanced Clinical Imaging Technology, Siemens Medical Solutions-CIBM, Lausanne, Switzerland, 2Radiology, UNIL, Lausanne, Switzerland, 3Signal Processing Laboratory (LTS5), EPFL, Lausanne, Switzerland, 4Mayo Clin, Rochester, MN USA, 5Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland

Normal aging and numerous diseases such as Alzheimer’s disease (AD), vascular dementia (VD) and other neurodegenerative diseases lead to brain tissue changes over time. In the interest of disease classification and diagnosis, it is highly desirable to have reliable and automatic tools to measure brain tissue volumes. In this study, we compare volumetric and GM probabilities differences extracted from standard T1-weighted images using SPM8, VBM8 and an in-house automatic tissue classification algorithm called VEMTC.

11:30 536.   Using multi-parametric quantitative MRI to model myelin in the brain 
J.B.M. Warntjes1,2, J. West1,3, O. Dahlqvist-Leinhard1,3, G. Helms4, A.-M. Landtblom5, and P. Lundberg6,7
1Linköping University, Center for Medical Image Science and Visualization, Linköping, Sweden, 2Department of Medicine and Health, Division of Clinical Physiology, Linköping, Sweden, 3Department of Medicine and Health, Division of Radiation Physics, Linköping, Sweden, 4University Medical Center, MR-Research in Neurology and Psychiatry, Göttingen, Germany, 5Department of Clinical Neuroscience, Linköping, Sweden, 6Linköping University, Dept of Radiation Physics and Dept of Radiology, IMH, University of Linkoping, Linköping, Sweden, 7University Hospital of Linköping, Dept of Radiation Physics and Dept of Radiology, CKOC, University Hospital of Linkoping, Linköping, Sweden

A model is proposed where myelin partial volume in brain parenchyma is estimated utilizing quantitative Magnetic Resonance Imaging. QMRI aims at the absolute measurement of physical parameters such as the relaxation rates R1 and R2 and proton density PD. Data on 9 brain structures of 30 healthy subjects were used to set the model parameters. The model estimated an average 30.6±1.2% myelin in healthy white matter. Examples are shown for clinical cases were both general and local reductions of myelin can be observed.

11:42 537.   Orthogonal Super Resolution Reconstruction for 3D Isotropic Imaging in 9.4T MRI 
Niranchana Manivannan1, Bradley D. Clymer1, Anna Bratasz2,3, and Kimerly A. Powell2,3
1Department Of Electrical and Computer Engineering, The Ohio State University, Columbus, Ohio, United States, 2Small Animal Imaging Shared Resource, The Ohio State University, 3Department of Biomedical Informatics, The Ohio State University, Columbus, Ohio, United States

The goal of this research is to apply orthogonal super resolution (SR) reconstruction technique to create isotropic 3D MRI images from 2D multislice stacks of images. The evaluation of this technique was performed in ex-vivo mouse model where the results of the SR reconstruction were quantitatively and qualitatively compared to an isotropically acquired 3D image . The structural detail observed in the through-plane direction of the SR reconstructed images was comparable to that observed in the isotropically acquired 3D scans. For the first time SR algorithm is successfully used to reconstruct in-vivo 3D isotropic volume in Ultra high field MRI.

11:54 538.   Addressing positioning induced variabillity in VBM analyses 
Costin Tanase1, Tyler Lesh1, and Cameron Carter1
1Psychiatry and Behavioral Sciences, University of California at Davis, Sacramento, CA, United States

VBM studies have suggested the presence of reduced gray matter (GM) in relatively focal lateral and medial prefrontal and temporal cortical regions that are present at the first episode schizophrenia and which appear to become more extensive with illness progression. However there are a number of methodological concerns that impact the interpretation of studies using the VBM approach. In this work we address two of the major sources of variability, such as the accuracy of normalization of GM to a standard template and the positioning in the scanner.

12:06 539.   Training-related cortical thickness changes 
Jan Scholz1, Miriam Klein2, and Heidi Johansen-Berg1
1University of Oxford, FMRIB Centre, Oxford, United Kingdom, 2University College London, Sobell Department of Motor Neuroscience and Movement Disorders, London

Evidence for training-related gray matter changes have been reported in several studies [1,2,3]. However, changes detected with gray matter VBM can potentially be due to global or local intensity changes, lesions, morphological changes, and/or thickness changes. Here we specifically test for changes in cortical thickness over time taking advantage of the longitudinal processing stream of FreeSurfer.

12:18 540.   A General-Purpuse Learning-Based Wrapper Method to Correct Systematic Errors in Automatic Image Segmentation: Consistently Improved Performance in Hippocampus, Cortex and Brain Segmentation 
Hongzhi Wang1, Sandhitsu R. Das1, Murat Altinay1, John Pluta1, Jung Wook Suh1, caryne craige1, Brian Avants1, and Paul Yushkevich1
1PICSL, Department of Radiology, University of Pennsylvania, Philadelphia, PA, United States

It is often a nontrivial task to produce optimal segmentation results using existing segmentation software, especially when the user's data and manual segmentation protocol are different from those used by the software developers. We present an open source wrapper algorithm that can automatically improve segmentation accuracy of any existing segmentation software on the user's data using training data provided by the user.