Ayman M. Khalifa

A collaborative resource to build consensus for automated left ventricular segmentation of cardiac MR images.

A collaborative framework was initiated to establish a community resource of ground truth segmentations from cardiac MRI. Multi-site, multi-vendor cardiac MRI datasets comprising 95 patients (73 men, 22 women; mean age 62.73±11.24years) with coronary artery disease and prior myocardial infarction, were randomly selected from data made available by the Cardiac Atlas Project (Fonseca et al., 2011). Three semi- and two fully-automated raters segmented the left ventricular myocardium from short-axis cardiac MR images as part of a challenge introduced at the STACOM 2011 MICCAI workshop (Suinesiaputra et al., 2012). Consensus myocardium images were generated based on the Expectation-Maximization principle implemented by the STAPLE algorithm (Warfield et al., 2004). The mean sensitivity, specificity, positive predictive and negative predictive values ranged between 0.63 and 0.85, 0.60 and 0.98, 0.56 and 0.94, and 0.83 and 0.92, respectively, against the STAPLE consensus. Spatial and temporal agreement varied in different amounts for each rater. STAPLE produced high quality consensus images if the region of interest was limited to the area of discrepancy between raters. To maintain the quality of the consensus, an objective measure based on the candidate automated rater performance distribution is proposed. The consensus segmentation based on a combination of manual and automated raters were more consistent than any particular rater, even those with manual input. The consensus is expected to improve with the addition of new automated contributions. This resource is open for future contributions, and is available as a test bed for the evaluation of new segmentation algorithms, through the Cardiac Atlas Project (www.cardiacatlas.org).

Myocardial segmentation using contour-constrained optical flow tracking

Despite the important role of object tracking using the Optical Flow (OF) in computer graphics applications, it has a limited role in segmenting speckle-free medical images such as magnetic resonance images of the heart. In this work, we propose a novel solution of the OF equation that allows incorporating additional constraints of the shape of the segmented object. We formulate a cost function that include the OF constraint in addition to myocardial contour properties such as smoothness and elasticity. The method is totally different from the common naive combination of OF estimation within the active contour model framework. The technique is applied to dataset of 20 patients and comparison with manual segmentation shows sensitivity and specificity levels of 93% and 99% respectively is obtained through the challenge validation system.

Performance of Optical Flow tracking approaches for cardiac motion analysis

Tagging Magnetic Resonance Imaging (MRI) sequence is used for evaluating Left Ventricular contractility. In this technique, a pattern of spatially varying magnetism is applied at the end diastole. Analyzing the deformation of tag pattern during the cardiac cycle has wide applications for cardiac deformation analysis. Noninvasive myocardial tagging in MRI has shown great potential in measuring and studying the motion of the heart. This paper presents a mathematical model that simulates the real cardiac motion during myocardial tagging. We synthesized both the Spatial Modulation of Magnetization (SPAMM) and complementary Spatial Modulation of Magnetization (CSPAMM) tag patterns with arbitrary spatial frequency. Using this model, we tested the performance and limitations of different Optical Flow (OF) motion tracking techniques and compared them with the performance of Harmonic Phase (HARP) analysis technique. The results exhibit that the OF tracking accuracy differs from point to another with a noticeable over estimation at the end of systole. Also OF is performing better than HARP at the heart borders.

Accurate estimation of the myocardium global function from reduced magnetic resonance image acquisitions.

Evaluating the heart global function from magnetic resonance images is based on estimating a number of functional parameters such as the left ventricular (LV) volume, LV mass, ejection fraction, and stroke volume. Estimating these parameters requires accurate calculation of the volumes enclosed by the inner and outer surfaces of the LV chamber at the max contraction and relaxation states of the heart. Currently, this is achieved through acquisition and segmentation of a large number of short-axis (SAX) views of the LV, which is time-consuming and expensive. Reducing the number of acquisitions results in undersampling the LV surfaces and hence increases the calculation errors. In this work, we describe and evaluate a method for estimating the cardiac parameters from a small number of image acquisitions that includes one long-axis (LAX) view of the LV. In this method, the LAX contour is used to swipe the SAX contours to fill in the missed LV surface between the SAX slices. Results on 25 patients and CT phantoms shows that, given the same number of slices, the proposed method is superior to other methods.

MRI T2* imaging for assessment of liver iron overload: study of different data analysis approaches

Background Recently, magnetic resonance imaging (MRI) has been established as an effective technique for evaluating iron overload by measuring T2* in the liver. Purpose To investigate the effects of various factors associated with T2* calculation on the resulting measurement and to determine the analysis criterion that provides the most accurate T2* measurements. Material and Methods Both phantom and in vivo MRI experiments were conducted to study the effects of the selected region of interest (ROI) location and size, signal-averaging method, exponential-fitting model, echo truncation, iron-overload severity, and inter-/intra-observer variabilities on T2* measurements. The results were compared to reference values from the scanner processing software. Results The pixel-by-pixel calculation method provided results in better agreement with the reference values from the MRI scanner than the average or median methods. The choice of the exponential fitting model affected the results, depending on signal-to-noise ratio, number of echoes, minimum and maximum echo times, and tissue composition inside the selected ROI. The single-exponential model resulted in smaller error than the bi-exponential or exponential-plus-constant models, where the latter two models showed similar results. The relative performance of the different models and methods was not affected by the degree of iron-overload. Conclusion Various factors associated with the adopted T2* calculation method affect the resulting measurement. In this study, the pixel-by-pixel calculation method and single-exponential model provided the most accurate results based on the conducted phantom and in vivo MRI experiments.

Evaluation of cardiac left ventricular mass from Tagged Magnetic Resonance Imaging

Tagged Magnetic Resonance Imaging (tMRI) is considered the gold standard for quantitative assessment of the cardiac regional function. However, quantification of the global function from tMRI is challenging due to the low contrast between myocardium and blood in these images, and hence, prevents the accurate segmentation of the myocardium. In this work, a method for enhancing the myocardium-to-blood contrast in tagged MR images is presented. First, the tag pattern in tMRI is removed by accurately suppressing the frequency components of the tag pattern to produce tagless images. Then, the image contrast is enhanced by estimating the local standard deviation and using it to suppress the blood signal intensity. The proposed method is applied on a dataset for 12 patients to calculate the global cardiac functional parameters. The results are then compared to those calculated from standard cine MRI image sequences of the same patients. The results show that the proposed method can be used to accurately estimate the left ventricular volume and mass.

Accurate analysis of cardiac tagged MRI using combined HARP and optical flow tracking

In this work, we present a new method for analyzing cardiac tagged Magnetic Resonance Imaging (tMRI). The method combines two major tracking techniques: Harmonic Phase (HARP) and Optical Flow (OF). The results of the two techniques are fused together to accurately estimate the displacement of each myocardium point. The developed methods were tested using numerical MRI phantom at different SNR levels and deformation rates. The results show that the proposed method is more accurate and reliable than the HARP and the OF methods.

Influence of the analysis technique on estimating hepatic iron content using MRI.

To investigate the effect of the analysis technique on estimating hepatic iron content using MRI.

Assessment of cardiac mass from tagged magnetic resonance images

PurposeTagged and cine magnetic resonance imaging (tMRI and cMRI) techniques are used for evaluating regional and global heart function, respectively. Measuring global function parameters directly from tMRI is challenging due to the obstruction of the anatomical structure by the tagging pattern. The purpose of this study was to develop a method for processing the tMRI images to improve the myocardium-blood contrast in order to estimate global function parameters from the processed images.Materials and methodsThe developed method consists of two stages: (1) removing the tagging pattern based on analyzing and modeling the signal distribution in the image’s k-space, and (2) enhancing the blood-myocardium contrast based on analyzing the signal intensity variability in the two tissues. The developed method is implemented on images from twelve human subjects.ResultsVentricular mass measured with the developed method showed good agreement with that measured from gold-standard cMRI images. Further, preliminary results on measuring ventricular volume using the developed method are presented.ConclusionThe promising results in this study show the potential of the developed method for evaluating both regional and global heart function from a single set of tMRI images, with associated reduction in scan time and patient discomfort.

Tag removal in cardiac tagged MRI images using coupled dictionary learning.

Tagged Magnetic Resonance Imaging (tMRI) is considered to be the gold standard for quantitative assessment of the cardiac local functions. However, the tagging patterns and low myocardium-to-blood-pool contrast of tagged images bring great challenges to cardiac image processing and analysis tasks such as myocardium segmentation and tracking. Hence, there has been growing interest in techniques for removing tagging lines. In this work, a method for removing tagging patterns in tagged MR images using a coupled dictionary learning (CDL) model is proposed. In this model, identical sparse representations are assumed for image patches in the tagged MRI and corresponding cine MRI image spaces. First, we learn a dictionary for the tagged MRI image space. Then, we compute a dictionary for the cine MRI image space so that corresponding tagged and cine patches have the same sparse codes in terms of their respective dictionaries. Finally, in order to produce the de-tagged (cine version) of a test tagged image, the sparse codes of the tagged patches and the trained cine dictionary are used together to construct the de-tagged patches. We have tested this tag removal method on a dataset of tagged cardiac MR images. Our experimental results compared favorably with a recently proposed tag removal method that removes tags in the frequency domain using an optimal band-stop filter of harmonic peaks.