Tagged MR Images Research Articles

Modern Mathematics in Biomedical Imaging

Over the past decades, modern mathematics is playing an increasingly important role in the development of biomedical imaging, yielding groundbreaking ideas and powerful tools for clinical and preclinical applications. Nevertheless, there are not only growing obstacles between applied mathematicians and imaging engineers to communicate more effectively but also great synergy to be unlocked across boundaries between mathematics and imaging technology. Suggested by the Editor-in-Chief, we organized this special issue to address theseunique challenges by presenting some sophisticated mathematical methods in an engineer-friendly language and demonstrating some representative applications of contemporary analytic tools in image reconstruction research. Nonrigid image registration and a diffeomorphic version of the demons algorithm have been commonly used to estimate tissue deformations in highly deformable anatomies. In the paper by G. Janssens et al. in this issue, the authors further developed the concept of continuous diffeomorphic flow and proposed a diffeomorphic version of the morphons registration method. This method can be used to obtain accurate estimation of deformations between images with variable contrast and hence can be applied to radiotherapy for lung cancer patients and to 4D respiratory-correlated CT of the thorax. Bioluminescence tomography (BLT) has been developed at a fast pace in recent years in study of physiological and pathological processes at cellular and molecular levels. In practice, fine discretization is required but may cause large datasets and an increase the ill-posedness of the problem. In the paper by J. Liu et al. in this issue, the authors present a multilevel sparse reconstruction method based on a framework of the finite-element method. Empirical results showed its effectiveness and potential applications in BLT. When one considers the time-integrated X-ray flux from multiple X-ray sources to shorten the data acquisition process, a promising way is to use overlapped projections from multiple X-ray sources. In order to perform image reconstruction effectively and efficiently from overlapping projections in this configuration, H. Yu et al. developed a multisource simultaneous-algebraic reconstruction technique regularized by a sparsity-oriented constraint in the soft-threshold filtering framework in their paper in this issue. Their numerical simulation further verified the proposed algorithm and demonstrated its advantages in image reconstruction from overlapping data. Mathematical modeling in dynamic positron emission tomography (PET) can be simplified, using compartment models as a linear system. To avoid invasive arterial sampling of blood to acquire values of tracer concentration, blind methods to estimate both blood input and kinetic parameters have recently drawn attention. In a paper by Y. Cheng and I. S. Yetik in this issue, the authors presented a method to bound errors in kinetic parameter estimates. Their findings are important to show the effect of the blood function on kinetic parameter estimation and to evaluate various blind methods. In 3D parallel-beam geometry of computed tomography, Y. Wei et al. demonstrated in a paper in this issue that the inverse Fourier transform in different coordinate systems leads to different reconstruction formulas. They explained the reason why the Radon formula cannot be applied to truncated projections. They then introduced a Γ coordinate system, computed its weight functions for the inverse Fourier transform, and obtained a simplified scanning model. Their results on the motion of frequency can be related to the classical Frenet-Serret theorem on 3D curves. In a paper by A. Becciu et al. in this issue, the authors developed a new formulation to detect critical points of 3D scalar images. Their approach was based on a topological number, which is the 3D generalization to three of the 2D winding numbers. They applied their new method to detection and counting of ovarian follicles and neuronal cells and to the estimation of cardiac motion from tagged MR images. The blood vessels and nerve trees consist of tubular objects interconnected into a complex, tree or web-like, geometric structure. Their large-scale range presents a measurement difficulty and an exponential increase of component numbers with decreasing scale. O. P. Dzyubak and E. L. Ritman proposed in a paper in this issue a solution for automation of an adaptive nonsupervised system for tracking tubular objects based on a multiscale framework and a Hessian-based object shape detector. We hope that researchers in applied mathematics and medical imaging will find this special issue interesting and useful. Yangbo Ye Robert J. Plemmons James G. Nagy Qinian Jin

3D left ventricular strain from unwrapped harmonic phase measurements

To validate a method for measuring 3D left ventricular (LV) strain from phase-unwrapped harmonic phase (HARP) images derived from tagged cardiac magnetic resonance imaging (MRI). A set of 40 human subjects were imaged with tagged MRI. In each study the HARP phase was computed and unwrapped in each short-axis and long-axis image. Inconsistencies in unwrapped phase were resolved using branch cuts manually placed with a graphical user interface. 3D strain maps were computed for all imaged timeframes in each study. The strain from unwrapped phase (SUP) and displacements were compared to those estimated by a feature-based (FB) technique and a HARP technique. 3D strain was computed in each timeframe through systole and mid-diastole in approximately 30 minutes per study. The standard deviation of the difference between strains measured by the FB and the SUP methods was less than 5% of the average of the strains from the two methods. The correlation between peak circumferential strain measured using the SUP and HARP techniques was over 83%. The SUP technique can reconstruct full 3D strain maps from tagged MR images through the cardiac cycle in a reasonable amount of time and user interaction compared to other 3D analysis methods.

Shortest path refinement for motion estimation from tagged MR images.

Magnetic resonance tagging makes it possible to measure the motion of tissues such as muscles in the heart and tongue. The harmonic phase (HARP) method largely automates the process of tracking points within tagged MR images, permitting many motion properties to be computed. However, HARP tracking can yield erroneous motion estimates due to 1) large deformations between image frames, 2) through-plane motion, and 3) tissue boundaries. Methods that incorporate the spatial continuity of motion--so-called refinement or flood-filling methods--have previously been reported to reduce tracking errors. This paper presents a new refinement method based on shortest path computations. The method uses a graph representation of the image and seeks an optimal tracking order from a specified seed to each point in the image by solving a single source shortest path problem. This minimizes the potential errors for those path dependent solutions that are found in other refinement methods. In addition to this, tracking in the presence of through-plane motion is improved by introducing synthetic tags at the reference time (when the tissue is not deformed). Experimental results on both tongue and cardiac images show that the proposed method can track the whole tissue more robustly and is also computationally efficient.

Tracking muscle deformation during speech from tagged and diffusion tensor magnetic resonance imaging.

Oral head and neck cancer is the sixth most common cancer worldwide. These tumors are usually treated by surgical removal of the affected tissue. The result of the surgery is a loss of muscle tissue, accompanied by scarring, reduced strength, and often reduced function. This paper explores the relationship between tongue muscle orientation and muscle deformation pattern in normal and post-cancer surgery speakers. Our motivation is the need to better understand the mechanisms that underlie tongue motion, in order to better interpret clinical observations and to provide data that can help predict optimal surgical outcomes. This study uses diffusion tensor imaging (DTI) to extract muscle fiber orientation, but it is not possible to image the muscle deformation in real-time using DTI. On the other hand, tagged MRI can track muscle fiber bundles in real time, but the original fiber bundle is not visible. Therefore, this work develops a method to overlay muscle bundles from the DTI onto the corresponding locations in tagged MR images and track muscle deformation. The mechanical properties of the muscles are calculated, such as translation and bending. Motion pattern differences between the normal and glossectomy tongue are detailed. [NIH R01-CA133015 and K99-DC9279.]

A Flow Quantification Method Using Fluid Dynamics Regularization and MR Tagging

This paper presents a new method for improved flow analysis and quantification using MRI. The method incorporates fluid dynamics to regularize the flow quantification from tagged MR images. Specifically, the flow quantification is formulated as a minimization problem based on the following: 1) the Navier-Stokes equation governing the fluid dynamics; 2) the flow continuity equation and boundary conditions; and 3) the data consistency constraint. The minimization is carried out using a genetic algorithm. This method is tested using both computer simulations and MR flow experiments. The results are evaluated using flow vector fields from the computational fluid dynamics software package as a reference, which show that the new method can achieve more realistic and accurate flow quantifications than the conventional method.

Texture Classification and Shape Statistics Variational Approach for Segmentation of Left Ventricle Tagged MR Images

分割带标记线核磁共振(tagged MR)图像是左心室运动重建的前提.由于标记线的加载破坏了左心室的轮廓边缘和区域灰度一致性,再加上乳突肌的存在,使带标记线核磁共振图像的左心室内外轮廓分割变得相当困难.在变分框架下,将纹理分类信息与形状统计先验知识引入Mumford-Shah模型中,提出了一种改进的分割带标记线核磁共振图像的左心室内外轮廓的方法.该方法基于支持向量机对S滤波器组提取的纹理特征的分类结果,构造了一种新的图像能量表示;针对乳突肌及边缘断裂现象,引入形状统计先验信息来约束曲线的演化.因为分割过程利用了有监督学习策略,较好地克服了标记线对左心室区域灰度的影响,提高了分割精度.实验结果表明,该方法较以往方法具有更高的分割精度和更好的稳定性.;Segmentation of left ventricle tagged MR images is the basis of ventricular motion reconstruction. In left ventricle tagged MR images, the boundaries are often obscured or corrupted by the tag lines and region inhomogeneity as well as the existence of papillary muscles. These factors increase the difficulty of segmenting the inner and outer contour of left ventricle precisely. This paper introduces texture classification information and shape statistical knowledge into the Mumford-Shah model and presents an improved texture classification and shape statistics variational approach for the segmentation of inner and outer contour of left ventricle. It uses the output of support vector machine (SVM) classifier relying on S filter banks to construct a new region-based image energy term. This approach can overcome the influence of tag lines because it makes use of the supervised classification strategy. The introduction of shape statistics can improve the segmentation with broken boundaries. Segmentation results of an entire cardiac period on an identical image layer and a comparison of mean absolute distance analysis between contours generated by this approach and that generated by hand demonstrate that this method can achieve a higher segmentation precision and a better stability than other various approaches.

Strain measurement in the left ventricle during systole with deformable image registration

The objective of this study was to validate a deformable image registration technique, termed Hyperelastic Warping, for left ventricular strain measurement during systole using cine-gated, non-tagged MR images with strains measured from tagged MRI. The technique combines deformation from high resolution, non-tagged MR image data with a detailed computational model, including estimated myocardial material properties, fiber direction, and active fiber contraction, to provide a comprehensive description of myocardial contractile function. A normal volunteer (male, age 30) with no history of cardiac pathology was imaged with a 1.5 T Siemens Avanto clinical scanner using a TrueFISP imaging sequence and a 32-channel cardiac coil. Both tagged and non-tagged cine MR images were obtained. The Hyperelastic Warping solution was evolved using a series of non-tagged images in ten intermediate phases from end-diastole to end-systole. The solution may be considered as ten separate warping problems with multiple templates and targets. At each stage, an active contraction was initially applied to a finite element model, and then image-based warping penalty forces were utilized to generate the final registration. Warping results for circumferential strain (R(2)=0.75) and radial strain (R(2)=0.78) were strongly correlated with results obtained from tagged MR images analyzed with a Harmonic Phase (HARP) algorithm. Results for fiber stretch, LV twist, and transmural strain distributions were in good agreement with experimental values in the literature. In conclusion, Hyperelastic Warping provides a unique alternative for quantifying regional LV deformation during systole without the need for tags.

Segmentation of cardiac cine-MR images and myocardial deformation assessment using level set methods

In this paper, we present an original method to assess the deformations of the left ventricular myocardium on cardiac cine-MRI. First, a segmentation process, based on a level set method is directly applied on a 2 D+ t dataset to detect endocardial contours. Second, the successive segmented contours are matched using a procedure of global alignment, followed by a morphing process based on a level set approach. Finally, local measurements of myocardial deformations are derived from the previously determined matched contours. The validation step is realized by comparing our results to the measurements achieved on the same patients by an expert using the semi-automated HARP reference method on tagged MR images.

LV Shape and Motion: B-Spline-Based Deformable Model and Sequential Motion Decomposition

In this paper, we extend a previous work by J. Park and propose a uniform framework to reconstruct left ventricle (LV) geometry/motion from tagged MR images. In our work, the LV is modeled as a generalized prolate spheroid, and its motion is decomposed into four components-global translation, polar radial/z-axis compression, twisting, and bending. By formulating model parameters as tensor products of B-splines, we develop efficient algorithms to quickly reconstruct LV geometry/motion from extracted boundary contours and tracked planar tags. Experiments on both synthesized and in vivo data are also reported.

Analysis of 3-D Myocardial Motion in Tagged MR Images Using Nonrigid Image Registration

Tagged magnetic resonance imaging (MRI) is unique in its ability to noninvasively image the motion and deformation of the heart in vivo, but one of the fundamental reasons limiting its use in the clinical environment is the absence of automated tools to derive clinically useful information from tagged MR images. In this paper, we present a novel and fully automated technique based on nonrigid image registration using multilevel free-form deformations (MFFDs) for the analysis of myocardial motion using tagged MRI. The novel aspect of our technique is its integrated nature for tag localization and deformation field reconstruction using image registration and voxel based similarity measures. To extract the motion field within the myocardium during systole we register a sequence of images taken during systole to a set of reference images taken at end-diastole, maximizing the normalized mutual information between the images. We use both short-axis and long-axis images of the heart to estimate the full four-dimensional motion field within the myocardium. We also present validation results from data acquired from twelve volunteers.

Quantitative analysis of left ventricular wall motion with MRI tagging

The purpose of this study was to establish a quantification of different parameters of left ventricular wall motion from tagged MR images. This evaluation method was to be applied to characterize the physiological contraction cycle and to determine pathophysiological changes. Myocardial tagging was performed at a basal, a mid-ventricular and an apical level of the left ventricle. A suitable software was programmed for the automatic quantification of rotation, contraction and circumferential shortening. The evaluation method was used in 8 healthy volunteers, 13 patients suffering from aortic stenosis before and one year after surgery and in 10 patients with myocardial infarction before and after revascularization. The software allows the quantification of left ventricular wall motion by assessment of rotation, contraction and circumferential shortening. In the healthy volunteers, there was a wringing motion with opposite rotation of base and apex of the heart. Before valve replacement, patients with aortic stenosis showed significantly increased apical rotation and torsion. One year after surgery, left-ventricular torsion had normalized. In patients with myocardial infarction, circumferential shortening increased after revascularization. The quantification of left ventricular wall motion using tagged MR images allows to characterize and follow-up changes of left ventricular wall motion in various diseases of the heart.

Level set based auto segmentation of the tagged left ventricle MR images.

To facilitate automatic segmentation, we adopt SVM (Support Vector Machine) to localize the left ventricle, and the segmentation is then carried out with narrow band level set. The method of generating the narrow band is improved such that the time used is reduced. Based on the imaging characteristics of the tagged left ventricle MR images, BPV (block-pixel variation) and intensity comparability are introduced to improve the speed term of level set and to increase the precision of segmentation. Our method can perform the segmentation of the tagged left ventricle MR images accurately and automatically.

Infarct Involution and Improved Function During Healing of Acute Myocardial Infarction: The Role of Microvascular Obstruction #,?

Delayed contrast-enhanced cardiac magnetic resonance imaging (ceCMR) delineates infarct size. The presence of hypoenhancement consistent with microvascular obstruction (MO) signifies larger infarcts with a worse prognosis. We hypothesized that the size of the contrast defect (CD) on ceCMR in acutely infarcted myocardium may change during infarct healing and depend upon the presence of MO. Twenty-five patients underwent CMR on weeks 1 and 8 after reperfused myocardial infarction. After short-axis cine CMR was performed, gadolinium was infused and ceCMR images and matched tagged cine MR images were obtained in the three most dysfunctional short-axis slices on cine CMR. The area and transmural extent of hyperenhancement (HE) with or without MO representing total CD size were planimetered. Between week 1 and week 8, the CD area fell from 1729+/-970 mm2 at week 1 to 1270+/-706 mm2 (p<0.001), as did the transmural extent of infarction (71+/-22% to 63+/-24%, p<0.001). The decline in CD trended to be higher in patients with MO (840+/-807 mm2) than in HE (312+/-485 mm2, p<0.07). In the patient group as a whole, ejection fraction (EF) improved (56+/-9% to 60+/-10%, p=0.002) between weeks 1 and 8, but patients with MO showed no increase in EF. Segments with some HE demonstrated partial functional improvement whereas no improvement was seen in HE+MO segments. In patients 8 weeks after reperfused myocardial infarction (MI), the size of infarction by ceCMR decreases compared to week 1 post-MI, especially in those with microvascular obstruction in whom there is little improvement in regional or global function.

Magnitude image CSPAMM reconstruction (MICSR).

Image reconstruction of tagged cardiac MR images using complementary spatial modulation of magnetization (CSPAMM) requires the subtraction of two complex datasets to remove the untagged signal. Although the resultant images typically have sharper and more persistent tags than images formed without complementary tagging pulses, handling the complex data is problematic and tag contrast still degrades significantly during diastole. This article presents a magnitude image CSPAMM reconstruction (MICSR) method that is simple to implement and produces images with improved contrast and tag persistence. The MICSR method uses only magnitude images - i.e., no complex data - but yields tags with zero mean, sinusoidal profiles. A trinary display of MICSR images emphasizes their long tag persistence and demonstrates a novel way to visualize myocardial deformation. MICSR contrast and contrast-to-noise ratios (CNR) were evaluated using simulations, a phantom, and two normal volunteers. Tag contrast 1000 msec after the R wave trigger was 3.0 times better with MICSR than with traditional CSPAMM reconstruction techniques, while CNRs were 2.0 times better.

Unsupervised reconstruction of a three-dimensional left ventricular strain from parallel tagged cardiac images.

A new algorithm, called the Unsupervised Tag ExTraction and Heart strain(E) Reconstruction (UNTETHER) algorithm, is presented for quantifying three-dimensional (3D) myocardial strain in tagged cardiac MR images. Five human volunteers and five postinfarct patients were imaged. 3D strains measured by UNTETHER and a user-supervised technique were compared. Each study was analyzed in 49 +/- 8 min with UNTETHER, compared to approximately 4 hr with the user-supervised technique. For pooled human data, the correlation coefficient between the two methods for circumferential shortening (E(cc)) was r = 0.91 at the mid-wall (P < 0.0005). UNTETHER is capable of measuring wall motion abnormalities resulting from coronary artery disease, and has the potential to overcome the main limitations (time and user-supervision requirements) to routine clinical use of tagged cardiac MRI.

Automated Model-Based Segmentation of the Left and Right Ventricles in Tagged Cardiac MRI.

We describe an automated, model-based method to segment the left and right ventricles in 4D tagged MR. We fit 3D epicardial and endocardial surface models to ventricle features we extract from the image data. Excellent segmentation is achieved using novel methods that (1) initialize the models and (2) that compute 3D model forces from 2D tagged MR images. The 3D forces guide the models to patient-specific anatomy while the fit is regularized via internal deformation strain energy of a thin plate. Deformation continues until the forces equilibrate or vanish. Validation of the segmentations is performed quantitatively and qualitatively on normal and diseased subjects.

High-resolution MRI of cardiac function with projection reconstruction and steady-state free precession.

The purpose of this study was to investigate the trabecular structure of the endocardial wall of the living human heart, and the effect of that structure on the measurement of myocardial function using MRI. High-resolution MR images (0.8 x 0.8 x 8 mm voxels) of cardiac function were obtained in five volunteers using a combination of undersampled projection reconstruction (PR) and steady-state free precession (SSFP) contrast in ECG-gated breath-held scans. These images provide movies of cardiac function with new levels of endocardial detail. The trabecular-papillary muscle complex, consisting of a mixture of blood and endocardial structures, is measured to constitute as much as 50% of the myocardial wall in some sectors. Myocardial wall strain measurements derived from tagged MR images show correlation between regions of trabeculae and papillary muscles and regions of high strain, leading to an overestimation of function in the lateral wall.

Tagged MRIによる脳動脈瘤モデルの血流動態解析

The purpose of this study was to investigate the possibility of visualization of hemodynamics in an intracranial aneurysm replica by using tagged MR imaging and particle image velocimetry (PIV) software. Our phantom system included a pulsatile flow pump and a replica of an anterior communicating artery aneurysm (Shelley Medical Imaging Technologies). Tagged MR imaging was carried out using a 1.5 T superconducting MR system and using ECG gated fast spoiled gradient echo sequence combined with spatial modulation of magnetization technique. Tag stripes were placed at 3mm intervals in a grid pattern in the aneurysm model. Analysis of tagged MR image data with PIV software revealed the velocity and direction of flowing fluid at each location in the aneurysm model every nine milliseconds. Tagged MR image velocimetry combined with tagged MR imaging and PIV analysis would be a promising and noninvasive technique for revealing hemodynamics in intracranial aneurysms.

Automated Segmentation of the Left and Right Ventricles in 4D Cardiac SPAMM Images.

In this paper we describe a completely automated volume-based method for the segmentation of the left and right ventricles in 4D tagged MR (SPAMM) images for quantitative cardiac analysis. We correct the background intensity variation in each volume caused by surface coils using a new scale-based fuzzy connectedness procedure. We apply 3D grayscale opening to the corrected data to create volumes containing only the blood filled regions. We threshold the volumes by minimizing region variance or by an adaptive statistical thresholding method. We isolate the ventricular blood filled regions using a novel approach based on spatial and temporal shape similarity. We use these regions to define the endocardium contours and use them to initialize an active contour that locates the epicardium through the gradient vector flow of an edgemap of a grayscale-closed image. Both quantitative and qualitative results on normal and diseased patients are presented.

Four-dimensional B-spline based motion analysis of tagged MR images:introduction and in vivo validation

In MRI tagging, magnetic tags - spatially encoded magneticsaturation planes - are created within tissues acting astemporary markers. Their deformation pattern provides usefulqualitative and quantitative information about the functionalproperties of underlying tissue and allows non-invasive analysisof mechanical function. The measured displacement at a given tagpoint contains only unidirectional information; in order to trackthe full 3D motion, these data have to be combined withinformation from other orthogonal tag sets over all time frames.Here, we provide a method to describe the motion of the heartusing a four-dimensional tensor product of B-splines. Invivo validation of this tracking algorithm is performed usingdifferent tagging sets on the same heart. Using the validationresults, the appropriate control point density was determined fornormal cardiac motion tracking. Since our motion fields areparametric and based on an image plane based Cartesiancoordinate system, trajectories or other derived values (velocity,acceleration, strains...) can be calculated for any desiredpoint within the volume spanned by the control points. Thismethod does not rely on specific chamber geometry, so the motionof any tagged structure can be tracked. Examples of displacementand strain analysis for both ventricles are also presented.