Electronic poster

Wednesday 13:30-15:30 Computer 38

13:30 3720. 4D Right Ventricular Strain in Pulmonary Hypertension and Normals

Bharath Ambale Venkatesh¹, Steven G. Lloyd², Mustafa I. Ahmed², Himanshu Gupta², Louid Dell'Italia², Thomas S. Denney Jr. <sup>1

1</sup>Electrical and Computer Engineering, Auburn University, Auburn, AL, United States; ²University of Alabama at Birmingham

Accurate assessment of right ventricular (RV) function is clinically important – particularly in patients with pulmonary hypertension (PHTN). Compared to the left ventricle (LV), however, analysis of RV function is relatively difficult because of relatively thin walls and lack of geometric symmetry. Also in PHTN, higher systolic blood pressure in the RV can cause excursion of the interventricular septum into the LV cavity causing it to lose its geometric symmetry. This abstract presents a method for reconstructing three-dimensional biventricular strain from tagged MRI in each imaged time frame through mid-diastole. This method is validated on normal volunteers and PHTN patients.

14:00 3721. A Novel Centerline Model for Cardiac Long Axis Wall Motion Analysis

Ting Song^1,2, Jeffrey A. Stainsby³, Maureen N. Hood^2,4, Vincent B. Ho<sup>2,4

1</sup>GE Healthcare Applied Science Laboratory, Rockville, MD, United States; ²Radiology, Uniformed Services University of the Health Sciences, Bethesda, MD, United States; ³GE Healthcare Applied Science Laboratory, Toronto, ON, Canada; ⁴Radiology, National Navy Medical Center, Bethesda, MD, United States

A novel long axis wall motion quantification model is proposed to provide a reliable and simple solution to cardiac function calculations. Using only routine clinical MR cine images, functional parameters can be quantified determined retrospectively.

14:30 3722. Automated Synchronization of Cardiac Phases for Myocardial BOLD MRI

Sotirios A. Tsaftaris^1,2, Xiangzhi Zhou², Richard Tang², Rachel Klein², Aggelos Katsaggelos¹, Rohan Dharmakumar<sup>2

1</sup>Electrical Engineering and Computer Science, Northwestern University, Evanston, IL, United States; ²Radiology, Northwestern University, Chicago, IL, United States

It is particularly important for the evaluation of cardiac phase-resolved myocardial blood-oxygen-level-dependent (BOLD) MRI studies, to robustly and reproducibly synchronize images from rest and stress studies. The possibility of visualizing BOLD signal changes in multiple cardiac phases is expected to increase the diagnostic confidence for identifying the affected myocardial territories. The purpose of this work is to develop automated statistical methods to facilitate in the robust and reproducible evaluation of cardiac phase-resolved myocardial BOLD MRI through temporal synchronization of rest and stress images acquired at different heart rates, without resorting to LV segmentation.

15:00 3723. Using Vector Velocity Imaging (VVI) to Measure Left Ventricular Systolic Strain and Diastolic Strain Rate in Cine MRI

Nicholas M. Dunn¹, Subha Raman², Helene Houle³, Gianni Pedrizzetti⁴, Mani Vannan², Orlando Simonetti<sup>2

1</sup>The Davis Heart and Lung Research Institute, The Ohio State University, Columbus, OH, United States; ²Internal Medicine, Division of Cardiovascular Medicine, The Ohio State University, Columbus, OH, United States; ³Siemens Medical Solutions, Mountain View, CA, United States; ⁴The University of Trieste, Trieste, Italy

Quantification of systolic strain and diastolic strain rate provide more sensitive indicators of myocardial dysfunction than subjective image interpretation. Vector Velocity Imaging (VVI) is a processing method developed to calculate myocardial strain and strain rate in echocardiography images. Its feasibility to calculate strain and strain rate in cine MR images was tested by analyzing short and long axis SSFP cine MR images of normal, healthy subjects and comparing the acquired values to literature values attained using other MR tissue-tagging methods. The results show that VVI may be used to calculate strain and strain rate in SSFP cine MRI.

Thursday 13:30-15:30 Computer 38

13:30 3724. Myocardial Infarction Segmentation by GMM Clustering Method with Free-Breathing 3D Navigator-Gated DE-MRI

Yonggang Lu^1,2, Thanh D. Nguyen³, Noel C. F. Codella³, Dorinna D. Mendoza⁴, Jonathan Weinsaft⁴, Bruce B. Lerman⁵, Yi Wang<sup>3

1</sup>Wiscom Intelligent System Co.,Ltd., Nanjing, Jiangsu, China; ²Department of Radiology , Weill Medical College of Cornell University, New York, NY, United States; ³Department of Radiology, Weill Medical College of Cornell University, New York, NY, United States; ⁴Department of Medicine, Division of Cardiology, Weill Medical College of Cornell University, New York, NY, United States; ⁵Department of Medicine, Weill Medical College of Cornell University, New York, NY, United States

A GMM clustering method with a free-breathing 3D Navigator-Gated DE-MRI was proposed for myocardial infarction segmentation in this study. Compared to commonly used methods, the novel method has a superior performance of more accuracy and operator-independence in assessing myocardial infarction as demonstrated by preliminary experiments with in vivo human data.

14:00 3725. Interstrain Comparisons of Murine Global Cardiac Mechanical Function Using MRI

Christakis Constantinides¹, Nikolas Aristokleous¹, Konstantinos Fokianos², Jeff Brandenburg³, Dimitrios Perperidis<sup>1

1</sup>Mechanical and Manufacturing Engineering, University of Cyprus, Nicosia, Cyprus; ²Mathematics and Statistics, University of Cyprus, Nicosia, Cyprus; ³Radiology, Duke University Medical Center, Durham, NC, United States

Quantitative characterization of ventricular function has become important for the assessment of cardiac performance in heart disease. As the manipulation of the mammalian genome becomes routine, it is now possible to generate animal models to study cardiovascular function and dysfunction. Critical to successful phenotypic screening of mouse models of the cardiovascular system using MRI are highly efficient four-dimensional (4D) acquisition protocols, and reduction of the computational image processing complexity for accurate quantification. The -goal of this study is the efficient, quantitative assessment of interstrain cardiac performance in C57BL/6J and DBA/2J mouse hearts under anesthesia, using MRI.

14:30 3726. Feasibility of Myocardial T1 Mapping from Cine-IR Images by Image Warping

Vincenzo Positano¹, Matteo Milanesi¹, Piergiorgio Masci¹, Thomas KIng Foo², J C. Hardy², Luca Marinelli², Andrea Barison, ^1,3, Daniele De Marchi¹, Massimo Lombardi¹, Luigi Landini<sup>4

1</sup>MRI Laboratory, "G- Monasterio" Foundation and Istitute of Clinical Physiology, Pisa, Italy; ²Global Research Center, General Electric, Niskayuna, NY, United States; ³Scuola Sant'Anna, Pisa, Italy; ⁴Department of Information Engineering, University of Pisa, Italy

Myocardial T1 mapping from Cine-IR images is feasible by warping the myocardium signal in each frame on a standardized model, evaluating the pixel-by-pixel T1 distribution on the model, and finally warping back the resulting T1 map on each frame.

15:00 3727. Myocardial Motion Estimation from Cardiac Cine-MRI with a Phase-Based Optical Flow Method

Marie Xavier¹, Alain Lalande¹, Paul Michael Walker¹, Jean-Christophe Eicher², Jean-Eric Wolf, ^1,2, François Brunotte¹, Louis Legrand<sup>1

1</sup>LE2I, University of Burgundy, Dijon, France; ²Department of Cardiology, University Hospital, Dijon, France

Generally, the evaluation of myocardial motion from cine-MRI sequences requires a visual evaluation of the regional contractile function and depends on the experience of the reader. To automatically detect local myocardial wall motion abnormalities from cine-MRI sequence, an optical flow technique based on phase information was used. First, the robustness of the technique with regards to Rician noise and to brightness variations was evaluated on synthetic images. Then, in the context of cardiac cine-MRI, a segmental decomposition of the myocardium allowed us to study the mean velocity along the cardiac cycle and gave similar values to those obtained by echocardiography.

Image Processing: Myocardium & Vascular

Hall B Monday 14:00-16:00 Computer 39

14:00 3728. Peak Angiogram Calculations from 4D Flow Imaging

Michael Loecher¹, Kevin Johnson¹, Christopher Francois², Oliver Wieben<sup>1

1</sup>Department of Medical Physics, University of Wisconsin, Madison, WI, United States; ²Department of Radiology, University of Wisconsin, Madison, WI, United States

This study aims to assess an alternative reconstruction method that utilizes the temporal information from a 4D radially encoded flow scan. The method creates and angiogram from dynamic time frames instead of a time averaged reconstruction. While the approach increases background noise, it alleviates the problem of signal drops and voids from reversing flow patterns. The utility of the algorithm was evaluated in a group of 4 volunteers and 6 patients, demonstrating improved signal consistency along the aorta.

14:30 3729. Unsupervised Reconstruction for Ungated Ghost Angiography by Clustering of Image Features

Sotirios A. Tsaftaris^1,2, Erik Offerman³, Robert R. Edelman³, Ioannis Koktzoglou<sup>3

1</sup>Electrical Engineering and Computer Science, Northwestern University, Evanston, IL, United States; ²Radiology, Northwestern University, Chicago, IL, United States; ³Radiology, NorthShore University HealthSystem, Evanston, IL, United States

Ghost magnetic resonance angiography (MRA) has been proposed as an unenhanced and ungated method for angiography. The method requires manual post-processing to identify suitable slices in a large stack from which to create an interpretable angiogram. To maximize the contrast of the final angiogram it is necessary to eliminate slices located within the body and to carefully select the slices that contain conspicuous ghost artifacts. This time-consuming process can also introduce unwanted inter- and intra- observer variability. The purpose of this work was to completely automate the reconstruction process during ungated and non-contrast-enhanced Ghost MRA using image analysis and clustering.

15:00 3730. Level-Set Segmentation of Arterial and Venous Vessels Based on ToF-SWI Data

Andreas Deistung¹, Michal Strzelecki², Andrzej Materka², Jürgen R. Reichenbach<sup>1

1</sup>Medical Physics Group, Department of Diagnostic and Interventional Radiology , Jena University Hospital , Jena, Germany; ²Institute of Electronics, Technical University of Lodz, Lodz, Poland

Non-invasive quantitative assessment of the cerebral vasculature is of high diagnostic and therapeutic interest. The pre-requisite for the quantitative description of blood vessels is voxel-wise classification into vessel and non-vessel structures. In this contribution, we use a hybrid level-set approach that relies on both boundary and region information to segment arterial and venous vessels from simultaneously acquired time-of-flight (ToF) and susceptibility weighted imaging (SWI) data to create a 3D representation of the arterial and venous vasculature.

15:30 3731. Fast Plaque Burden Assessment of the Femoral Artery Using 3D Black-Blood MRI and Automated Segmentation

Bernard Chiu¹, Xihai Zhao¹, Jinnan Wang², Niranjan Balu¹, Chun Yuan¹, William S. Kerwin<sup>1

1</sup>Radiology, University of Washington, Seattle, WA, United States; ²Clinical Sites Research Program, Philips Research North America, Briarcliff Manor, NY, United States

Peripheral arterial disease (PAD) is a serious health issue in the western world. Recent advances in high-resolution MRI have allowed noninvasive and detailed assessment of PAD, including black-blood MRI visualization of the vessel wall. Because the length of a femoral artery is substantial, a long field of view is required to image the femoral artery. Manual outlining of wall boundaries along the entire length of the femoral artery is an arduous task. In this work, we proposed and demonstrated an automatic algorithm that is capable of accurately segmenting the lumen and wall boundaries along the majority of the femoral artery.

Tuesday 13:30-15:30 Computer 39

13:30 3732. Hemorrhage Delineation and Blood Suppression Evaluation in Slab-Selection Phase-Sensitive Inversion-Recovery (SPI) Sequence with MRI

Dongxiang Xu¹, Jinnan Wang², Williams Kerwin¹, Chun Yuan<sup>1

1</sup>Radiology, University of Washington, Seattle, WA, United States; ²Philips Research North America, [e-mail], Seattle, WA, United States

Intraplaque hemorrhage (IPH) into the carotid atherosclerotic plaque has shown significant association with clinical symptoms and is believed to be a major factor causing plaque instability and progression according to previous histopathological and prospective studies. With the development of magnetic resonance imaging (MRI) in clinical diagnostics, several techniques have been developed to enable and improve the IPH evaluation. However, automated hemorrhage detection has been challenging due to either low IPH contrast or poor lumen contrast. Purpose: In this study, by incorporating the improved IPH and lumen contrasts in Slab-selection Phase-sensitive Inversion-recovery (SPI) MRI sequence, we develop a novel and robust image segmentation approach to automatically locate and delineate IPH in MR data. Quantitative IPH and lumen analysis results by this automatic segmentation technique were compared to a human reader, which demonstrated highly consistent performance.

14:00 3733. Direct and Indirect Surface Coil Correction for Cardiac Perfusion MRI

Hui Xue¹, Sven Zuehlsdorff², Jens Guehring<sup>1

1</sup>Corporate Research, Siemens Corporation, Princeton, NJ, United States; ²CMR Research and Development, Siemens Healthcare, Chicago, IL, United States

Although the first-pass myocardial perfusion MRI has proven its effectiveness in the early diagnosis of suspected ischemic heart diseases, this technique is still not routinely used. Certain technical difficulties prevent perfusion MRI from being added into the clinical workflow. Among of them includes the B1-field inhomogeneity caused by non-uniform characteristics of the receiver coils which still lacks intensive studies, when compared to perfusion imaging sequences or motion compensation. We therefore propose algorithms to perform the surface coil inhomogeneity correction (SCC) using proton density (PD) weighted images and B-Spline Free-Form Deformation (FFD).

14:30 3734. Saturation Correction of Dynamic Contrast Enhanced MRI Uptake Curves for Quantitative Myocardial Blood Flow Measurements Using an Assumed T₁ for Blood

John David Biglands¹, Abdulghani Larghat¹, Sven Plein¹, David L. Buckley¹, Michael Jerosch-Herold², Derek Magee³, Roger Boyle³, Aleksandra Radjenovic<sup>1

1</sup>School of Medicine, University of Leeds, Leeds, UK, United Kingdom; ²Radiology, Brigham and Womens Hospital, Boston, MA, United States; ³School of Computing, University of Leeds, Leeds, United Kingdom

Dynamic contrast enhanced magnetic resonance imaging of the myocardium using sufficiently high doses to be clinically useful generates uptake curves that require correction for signal saturation effects before they can be used for myocardial blood flow (MBF) estimation. Such corrections require knowledge of the native T1 of the blood and myocardium. This abstract shows that using an assumed blood T1 enables saturation correction of typical clinical datasets without the need for time consuming T1 measurements. MBF estimates from nine patients were consistent with literature values and were shown to be robust to variations in the assumed T1 of blood.

15:00 3735. Novel MRI T2 Mapping for Improved Myocardial Tissue Characterisation

Taigang He¹, Sanjay Prasad¹, Guang-Zhong Yang¹, Dudley Pennell¹, David Firmin<sup>1

1</sup>Royal Brompton Hospital and Imperial College London, London, United Kingdom

Synopsis: The aim of this study was to develop a novel T2 mapping method for improved myocardial tissue characterisation. The developed T2 sequence resulted in improved resolution with shorter echo time and echo spacing. The novel technique was consequently evaluated on nine human subjects. Preliminary results demonstrated that all images acquired were of good quality. Pixel wise T2 curve is well fitted and T2 mapping in the whole myocardium appeared homogeneous. This study suggests that T2 mapping may potentially be used for assessing regional disease variations across the myocardium.

Wednesday 13:30-15:30 Computer 39

13:30 3736. Cardiac Diffusion MR Microscopy of Rabbit Heart

Min Sig Hwang¹, Katja Odening², Ohad Ziv², Bum-Rak Choi², Gideon Koren², John R. Forder<sup>1

1</sup>McKnight Brain Institute, University of Florida, Gainesville, FL, United States; ²Cardiovascular Research Center, Rhode Island Hospital Alert Medical School of Brown University, Providence, RI, United States

In this study, we explored the potential of microscopic high angular resolution diffusion imaging (MHARDI) achieving a cellular level spatial resolution as a non-invasive tool that is sensitive to subtle changes in the heterogeneous microstructure and arrangement of the cardiac tissues. Diffusion tensor images and tensor invariants acquired with two diffusion sensitizing factors were investigated. Our results suggest that MHARDI with an optimized b-value and resolution may be a powerful tool for non-invasive monitoring of electro-mechanical property and its well-coordinated function.

14:00 3737. Automated Segmentation of Left Ventricle in Cine Cardiac MR Images: Experience from a Large Study

YingLi Lu¹, Perry Radau¹, Kim A. Connelly^1,2, Alexander Dick³, Graham A. Wright<sup>1

1</sup>Imaging Research, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; ²Cardiology, St Michael's Hospital, Toronto, ON, Canada; ³Cardiology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada

Purpose of this study is to develop a fully automatic left ventricle segmentation method from cine short-axis MR images and evaluate it on a large data set of 147 subjects grouped by pathology. Advantages of this method include that it: 1) does not require manually drawn contours; 2) provides not only endocardial and epicardial contours, but also papillary muscles and trabeculations¡¯ contours; 3) introduces a roundness measure that automatically locates the left ventricle; 4) simplifies the epicardial contour segmentation by mapping the pixels from Cartesian to approximately polar coordinates.

14:30 3738. Three-Dimensional Myocardial Tissue Tracking and Strain Calculation for Volumetric Cine DENSE Data

Xiaodong Zhong^1,2, Bruce S. Spottiswoode³, Craig H. Meyer^2,4, Frederick H. Epstein<sup>2,4

1</sup>MR R&D Collaborations, Siemens Healthcare, Atlanta, GA, United States; ²Biomedical Engineering, University of Virginia, Charlottesville, VA, United States; ³MRC/UCT Medical Imaging Research Unit, University of Cape Town, Cape Town, Western Cape, South Africa; ⁴Radiology, University of Virginia, Charlottesville, VA, United States

This abstract introduces novel automatic algorithms for myocardial tissue tracking and strain calculation for three-dimensional (3D) cine DENSE data. Specifically, scattered data interpolation using radial basis functions (RBF) was developed for Lagrangian tissue tracking. Also, a finite-strain based algorithm was developed to calculate the deformation gradient tensor and the Lagrangian strain tensor. The algorithms were performed on 3D cine DENSE data from five healthy volunteers to obtain 3D Lagrangian displacement and strain fields. The 3D myocardial mechanics, including normal strains, twist and torsion, were consistent with previous results from myocardial tagging in healthy volunteers.

15:00 3739. Towards Non-Invasive Automatic Detection of Cardiac Pathology by Strain and Rotation Analysis

Hans C. van Assen¹, Luc M.J. Florack², Frank F.J. Simonis¹, Jos J.M. Westenberg³, Gustav J. Strijkers¹, Bart M. ter Haar Romeny<sup>1

1</sup>Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Noord Brabant, Netherlands; ²Mathematics and Computer Science, Eindhoven University of Technology, Eindhoven, Noord Brabant, Netherlands; ³Radiology, Leiden University Medical Center, Leiden, Netherlands

This paper describes a novel image processing method for automated detection of cardiac pathology. It entails tagging analysis by means of an optical flow approach. Tag fading is overcome by exploitation of tag phase - retrieved by Gabor filtering - instead of tag brightness.

The method yields both the motion field and its first order derivative structure, necessary to calculate strain and rotation. Calculation of these derived parameters thus becomes straightforward. High-resolution in-slice cardiac strain and rotation are presented for four volunteers and a patient, and clearly show deviations for a patient with known small infarctions and wall motion abnormalities.