Application of automatic boundary detection for computerized quantitative analysis of left ventricular regional wall motion by two-dimensional echocardiography

This study was designed to set up a computer-aided image processing algorithm for boundary detection from two-dimensional echocardiography and to establish a computerized model for quantitative analysis of left ventricular wall motion with the application of automatic boundary detection. The four-chamber view echocardiographic images of seven normal subjects and five patients with acute myocardial infarction were investigated. The main steps of image processing in this algorithm included automatic threshold estimation, contrast stretching, radial search of endocardial boundary, and smoothing of the boundary. The displacements of the left ventricular endocardial contour from end-diastolic to end-systolic frame were measured using a sample point connection model. For analysis of the regional contraction, the left ventricular endocardial contour was divided equally into six segments. The wall motion curves in patients were compared with the normal wall motion pattern established from the normal subjects to identify the segments with normal or abnormal wall motion. The results of this quantitative method were compared with those of qualitative analysis. In the 30 segments of the five patients, quantitative analysis correctly identified nine of the 11 segments with abnormal wall motion diagnosed by qualitative analysis (sensitivity, 82%) and identified 17 of the 19 segments with normal wall motion (specificity, 89%). The positive and negative predictive values of quantitative analysis were 82% (9 of 11) and 89% (17 of 19), respectively, and the diagnostic accuracy was 87% (26 of 30). Our preliminary results suggest that computer-aided boundary detection can be applied to establish an objective and useful model for quantitative analysis of left ventricular regional wall motion.     

Segmentation of fetal ultrasound images

This paper describes a new method for segmentation of fetal anatomic structures from echographic images. More specifically, we estimate and measure the contours of the femur and of cranial cross-sections of fetal bodies, which can thus be automatically measured. Contour estimation is formulated as a statistical estimation problem, where both the contour and the observation model parameters are unknown. The observation model (or likelihood function) relates, in probabilistic terms, the observed image with the underlying contour. This likelihood function is derived from a region-based statistical image model. The contour and the observation model parameters are estimated according to the maximum likelihood (ML) criterion, via deterministic iterative algorithms. Experiments reported in the paper, using synthetic and real images, testify for the adequacy and good performance of the proposed approach.     

Noninvasive coronary angiography using computed tomography

While noninvasive imaging of the coronary lumen remains challenging, great strides have been made with computed tomography. Two variations of computed tomography are used in the study of the coronary tree: multislice or multidetector computed tomography and electron-beam computed tomography. Both have high spatial and temporal resolutions as well as excellent signal-to-noise ratios, which allows major branches of the coronary tree to be depicted. Impaired image quality, due to dense calcifications and multiple image artifacts including coronary artery motion and breathing artifacts, limits the clinical utility of noninvasive coronary angiography. Early studies with electron-beam angiography demonstrated an overall sensitivity of 85% and specificity of 89% for the detection of obstructive coronary artery disease. With early diastolic imaging, the sensitivity and specificity increases to 92 and 93%, respectively (rather than 80% of the cardiac interbeat interval, where coronary motion is more pronounced). Multidetector computed tomography, with improved spatial resolution but decreased temporal resolution, produces results that vary depending on the equipment. Four-slice scanners have an average sensitivity of only 61%, and only 38% of patients have all four vessels or 15 segments available for analysis, due to both cardiac motion and calcification. Thinner slice collimation with eight and 16 slices have allowed for improved detection. Sensitivity and specificity improve to 80 and 86%, respectively. Furthermore, the number of assessable segments with eight-2011; to 16-2011;slice scanners improves significantly, compared with four-slice scanners (85 vs. 73%; p < 0.001). If only assessable segments are included in analysis, sensitivity and specificity for multidetector-row computed tomograpy improves to nearly 90%. Compared with magnetic resonance imaging, with a reported accuracy of 72% in the only multicenter study, computed tomography has great promise to become the primary method of noninvasive coronary angiography.     

Noninvasive coronary angiography using computed tomography

While noninvasive imaging of the coronary lumen remains challenging, great strides have been made with computed tomography. Two variations of computed tomography are used in the study of the coronary tree: multislice or multidetector computed tomography and electron-beam computed tomography. Both have high spatial and temporal resolutions as well as excellent signal-to-noise ratios, which allows major branches of the coronary tree to be depicted. Impaired image quality, due to dense calcifications and multiple image artifacts including coronary artery motion and breathing artifacts, limits the clinical utility of noninvasive coronary angiography. Early studies with electron-beam angiography demonstrated an overall sensitivity of 85% and specificity of 89% for the detection of obstructive coronary artery disease. With early diastolic imaging, the sensitivity and specificity increases to 92 and 93%, respectively (rather than 80% of the cardiac interbeat interval, where coronary motion is more pronounced). Multidetector computed tomography, with improved spatial resolution but decreased temporal resolution, produces results that vary depending on the equipment. Four-slice scanners have an average sensitivity of only 61%, and only 38% of patients have all four vessels or 15 segments available for analysis, due to both cardiac motion and calcification. Thinner slice collimation with eight and 16 slices have allowed for improved detection. Sensitivity and specificity improve to 80 and 86%, respectively. Furthermore, the number of assessable segments with eight-to 16-slice scanners improves significantly, compared with four-slice scanners (85 vs. 73%; p<0.001). If only assessable segments are included in analysis, sensitivity and specificity for multidetector-row computed tomography improves to nearly 90%. Compared with magnetic resonance imaging, with a reported accuracy of 72% in the only multicenter study, computed tomography has great promise to become the primary method of noninvasive coronary angiography.     

A variational change detection method for multitemporal SAR images

In this letter, we develop a variational model for change detection in multitemporal synthetic aperture radar (SAR) images. SAR images are typically polluted by multiplicative noise, therefore ordinary active contour model (ACM), or the snake model, for image segmentation is not suitable for change detection in multitemporal SAR images. Our model is a generalization of ACM under the assumption that the image data fits the Generalized Gaussian Mixture (GGM) model. Our method first computes the log-ratio image of the input multitemporal SAR images. Then the method iteratively executes the following two steps until convergence: (1) estimate the parameters for the generalized Gaussian distributions inside and outside the current evolving curve using maximum-likelihood estimation; (2) evolve the current curve according to the image data and the parameters previously estimated. When convergence is achieved, the location of the evolving curve depicts the changed and the unchanged areas.

Experiments were carried out on both semi-simulated data set and real data set. Results showed that the proposed method achieves total error rates of 0.43% and 1.05%, for semi-simulated and real data sets, respectively, which were comparable to other prevalent methods.     

Automatic model-based contour detection of left ventricle myocardium from cardiac CT images

Purpose

For accurate evaluation of myocardial perfusion on computed tomography images, precise identification of the myocardial borders of the left ventricle (LV) is mandatory. In this article, we propose a method to detect the contour of LV myocardium automatically and accurately.

Methods

Our detection method is based on active shape model. For precise detection, we estimate the pose and shape parameters separately by three steps: LV coordinate system estimation, myocardial shape estimation, and transformation. In LV coordinate system estimation, we detect heart features followed by the entire LV by introducing machine-learning approach. Since the combination of two types feature detection covers the LV variation, such as pose or shape, we can estimate the LV coordinate system robustly. In myocardial shape estimation, we minimize the energy function including pattern error around myocardium with adjustment of pattern model to input image using estimated concentration of contrast dye. Finally, we detect LV myocardial contours in the input images by transforming the estimated myocardial shape using the matrix composed of the vectors calculated by the LV coordinate system estimation.

Results

In our experiments with 211 images from 145 patients, mean myocardial contours point-to-point errors for our method as compared to ground truth were 1.02 mm for LV endocardium and 1.07 mm for LV epicardium. The average computation time was 2.4 s (on a 3.46 GHz processor with 2-multithreading process).

Conclusions

Our method achieved accurate and fast myocardial contour detection which may be sufficient for myocardial perfusion examination.     

Automatic X-ray landmark detection and shape segmentation via data-driven joint estimation of image displacements

In this paper, we propose a new method for fully-automatic landmark detection and shape segmentation in X-ray images. To detect landmarks, we estimate the displacements from some randomly sampled image patches to the (unknown) landmark positions, and then we integrate these predictions via a voting scheme. Our key contribution is a new algorithm for estimating these displacements. Different from other methods where each image patch independently predicts its displacement, we jointly estimate the displacements from all patches together in a data driven way, by considering not only the training data but also geometric constraints on the test image. The displacements estimation is formulated as a convex optimization problem that can be solved efficiently. Finally, we use the sparse shape composition model as the a priori information to regularize the landmark positions and thus generate the segmented shape contour. We validate our method on X-ray image datasets of three different anatomical structures: complete femur, proximal femur and pelvis. Experiments show that our method is accurate and robust in landmark detection, and, combined with the shape model, gives a better or comparable performance in shape segmentation compared to state-of-the art methods. Finally, a preliminary study using CT data shows the extensibility of our method to 3D data.     

Quantitative measurements in IVUS images

IntraVascular UltraSound (IVUS) is a catheter-based technique which provides real-time high resolution tomographic images of both the lumen and arterial wall of a coronary segment, this in contrast to X-ray arteriography that provides a shadow image (luminogram) of the entire lumen. Nowadays the lumen and vessel parameters are measured manually, which is very time consuming and suffers from high inter- and intra-obser variability. With the continuing improvement in IVUS imaging, it is now feasible to develop and clinically apply automated methods of three-dimensional quantitative analysis of the coronary vessel morphology in an objective and reproducible way with automated contour detection techniques (QCU). Quantification, in 2D and 3D, as well as volume rendering for visualization of the IVUS images requires segmentation of the images (contour detection). The 3D contour detection system described in this article is based on the combination of contour detection in the transversal and sagital view. This article provides some of the basic principles of IVUS, the IVUS image quantification, the three-dimensional reconstruction and the contour detection and quantification in three-dimensional IVUS images.     

Statistical Region-Based Segmentation of Ultrasound Images

Segmentation of ultrasound images is a challenging problem due to speckle, which corrupts the image and can result in weak or missing image boundaries, poor signal to noise ratio and diminished contrast resolution. Speckle is a random interference pattern that is characterized by an asymmetric distribution as well as significant spatial correlation. These attributes of speckle are challenging to model in a segmentation approach, so many previous ultrasound segmentation methods simplify the problem by assuming that the speckle is white and/or Gaussian distributed. Unlike these methods, in this article we present an ultrasound-specific segmentation approach that addresses both the spatial correlation of the data as well as its intensity distribution. We first decorrelate the image and then apply a region-based active contour whose motion is derived from an appropriate parametric distribution for maximum likelihood image segmentation. We consider zero-mean complex Gaussian, Rayleigh, and Fisher-Tippett flows, which are designed to model fully formed speckle in the in-phase/quadrature (IQ), envelope detected, and display (log compressed) images, respectively. We present experimental results demonstrating the effectiveness of our method and compare the results with other parametric and nonparametric active contours. (E-mail:greg.slabaugh@gmail.com)     

River detection in high-resolution SAR data using the Frangi filter and shearlet features

ABSTRACT

River extraction plays an important role in several applications such as monitoring and navigation, and synthetic aperture radar (SAR) is one of the major sensors of remote sensing. This paper proposes an algorithm to detect a river from high-resolution SAR images mainly based on the Frangi filter and shearlet features with the help of an active contour model (ACM). The Frangi filter is firstly applied to enhance the river and then the shearlet features are computed by the shearlet transform. A rule of feature selection is then proposed to acquire the corresponding features of the river. Finally, binarization and an active contour model are implemented to extract the river. The approach is tested on SAR images and the experimental results demonstrate that the proposed method is effective.     

Towards quantitative analysis of coronary CTA

The current high spatial and temporal resolution, multi-slice imaging capability, and ECG-gated reconstruction of multi-slice computed tomography (MSCT) allows the non-invasive 3D imaging of opacified coronary arteries. MSCT coronary angiography studies are currently carried out by the visual inspection of the degree of stenosis and it has been shown that the assessment with sensitivities and specificities of 90% and higher can be achieved. To increase the reproducibility of the analysis, we present a method that performs the quantitative analysis of coronary artery diseases with limited user interaction: only the positioning of one or two seed points is required. The method allows the segmentation of the entire left or right coronary tree by the positioning of a single seed point, and an extensive evaluation of a particular vessel segment by placing a proximal and distal seed point. The presented method consists of: (1) the segmentation of the coronary vessels, (2) the extraction of the vessel centerline, (3) the reformatting of the image volume, (4) a combination of longitudinal and transversal contour detection, and (5) the quantification of vessel morphological parameters. The method is illustrated in this paper by the segmentation of the left and right coronary trees and by the analysis of a coronary artery segment. The sensitivity of the positioning of the seed points is studied by varying the position of the proximal and distal seed points with a standard deviation of 6 and 8 mm (along the vessels course) respectively. It is shown that only close to the individual seed points the vessel centerlines deviate and that for more than 80% of the centerlines the paths coincide. Since the quantification depends on the determination of the centerline, no user variability is expected as long as the seed points are positioned reasonably far away from the vessel lesion. The major bottleneck of MSCT imaging of the coronary arteries is the potential lack of image quality due to limitations in the spatial and temporal resolution, irregular or high heart beat, respiratory effects, and variations of the distribution of the contrast agent: the number of rejected vessel segments in diagnostic studies is currently still too high for implementation in routine clinical practice. Also for the automated quantitative analysis of the coronary arteries high image quality is required. However, based upon the trend in technological development of MSCT scanners, there is no doubt that the quantitative analysis of MSCT coronary angiography will benefit from these technological advances in the near future.     

Estimation of 3D left ventricular deformation from echocardiography

The quantitative estimation of regional cardiac deformation from 3D image sequences has important clinical implications for the assessment of viability in the heart wall. Such estimates have so far been obtained almost exclusively from Magnetic Resonance (MR) images, specifically MR tagging. In this paper we describe a methodology for estimating cardiac deformations from 3D echocardiography (3DE). The images are segmented interactively and then initial correspondence is established using a shape-tracking approach. A dense motion field is then estimated using a transversely isotropic linear elastic model, which accounts for the fiber directions in the left ventricle. The dense motion field is in turn used to calculate the deformation of the heart wall in terms of strain in cardiac specific directions. The strains obtained using this approach in open-chest dogs before and after coronary occlusion, show good agreement with previously published results in the literature. They also exhibit a high correlation with strains produced in the same animals using implanted sonomicrometers. This proposed method provides quantitative regional 3D estimates of heart deformation from ultrasound images.     

Differentiation of myocardial infarction and angina pectoris by processing ultrasonic color kinesis images

Purpose The aim of this study was to develop a method for early, accurate differentiation between old myocardial infarction (OMI) and angina pectoris (AP) using color kinesis (CK) images. We first extracted exact end-diastolic and end-systolic contours from CK images and then extracted the features of cardiac function from two CK images (one at rest, the other after exercise) and investigated their effectiveness in differentiating old myocardial infarction and angina pectoris. We then evaluated the effectiveness of several features in recognizing coronary artery disease and used the effective features to show the differentiation results.Methods First, we extracted exact end-diastolic and end-systolic contours from CK images with an active contour model. Second, we defined the features that seemed to be effective in recognizing coronary artery disease. The features are extracted from the region between the end-diastolic endocardial contour and end-systolic endocardial contour in two CK images: one obtained when the subject was at rest and the other after exercise. Nine features were considered effective for differentiating old myocardial infarction and angina pectoris, and the effectiveness in recognizing coronary artery disease, which includes old myocardial infarction and angina pectoris, was evaluated. Third, coronary artery disease is recognized by the effective features.Results Contours near a manual trace by a skilled physician were obtained using the proposed method. Multiple comparisons of the mean values of the extracted features were drawn among three groups: a healthy-subject group; an old myocardial infarction patient group; and an angina pectoris patient group. The feature effective in differentiating old myocardial infarction was the “area at rest”; those effective in differentiating angina pectoris were a “decrease in area” and a “decrease in movement.” These effective features have almost always differentiated old myocardial infarction and angina pectoris.Conclusions This study used the endocardial contour extraction technique with the dynamic contour model and evaluated the validity of the features of cardiac function; it then recognized coronary artery disease from the effective features. Multiple comparisons of the mean value of the extracted features among the healthy-subject group, the old myocardial infarction patient group, and the angina pectoris patient group has proved that the “area at rest” is effective in differentiating old myocardial infarction, and the “decrease in area” and “decrease in movement” are effective for differentiating angina pectoris.     

CT冠状动脉血管造影和血管内超声融合算法的研究

提出了一种基于冠状动脉中心线的CT冠状动脉血管造影(CTCA)和血管内超声(IVUS)融合的方法。该方法,首先从CTCA中提取冠状动脉中心线,并沿中心线提取其横截面;其次,对IVUS内膜和中-外膜进行半自动分割;再次,使用环形最大强度投影算法把CTCA和IVUS投影到二维空间中,并使用投影图像进行配准。最后,使用图形学方法把CTCA和IVUS融合在同一个三维空间中。融合CTCA和IVUS可以给医生诊断冠心病提供更多有用的信息。     

Image-based cardiac gating for three-dimensional intravascular ultrasound imaging

Three-dimensional (3-D) intravascular ultrasound (US), or IVUS, provides valuable insight into the tissue characteristics of the coronary wall and plaque composition. However, artefacts due to cardiac motion and vessel wall pulsation limit the accuracy and variability of coronary lumen and plaque volume measurement in 3-D IVUS images. ECG-gated image acquisition can reduce these artefacts but it requires recording the ECG signal and may increase image acquisition time. The goal of our study was to reconstruct a 3-D IVUS image with negligible cardiac motion and vessel pulsation artefacts, by developing an image-based gating method to track 2-D IVUS images over the cardiac cycle. Our approach involved selecting 2-D IVUS images belonging to the same cardiac phase from an asynchronously-acquired series, by tracking the changing lumen contour over the cardiac cycle. The algorithm was tested with IVUS images of a custom-built coronary vessel phantom and with patient images. The artefact reduction achieved using the image-gating approach was > 86% in the in vitro images and > 80% in the in vivo images in our study. Our study shows that image-based gating of IVUS images provides a useful method for accurate reconstruction of 3-D IVUS images with reduced cardiac motion artefact.     

Reversible jump MCMC methods for fully automatic motion analysis in tagged MRI

Tagged magnetic resonance imaging (tMRI) is a well-known noninvasive method for studying regional heart dynamics. It offers great potential for quantitative analysis of a variety of kine(ma)tic parameters, but its clinical use has so far been limited, in part due to the lack of robustness and accuracy of existing tag tracking algorithms in dealing with low (and intrinsically time-varying) image quality. In this paper, we evaluate the performance of four frequently used concepts found in the literature (optical flow, harmonic phase (HARP) magnetic resonance imaging, active contour fitting, and non-rigid image registration) for cardiac motion analysis in 2D tMRI image sequences, using both synthetic image data (with ground truth) and real data from preclinical (small animal) and clinical (human) studies. In addition we propose a new probabilistic method for tag tracking that serves as a complementary step to existing methods. The new method is based on a Bayesian estimation framework, implemented by means of reversible jump Markov chain Monte Carlo (MCMC) methods, and combines information about the heart dynamics, the imaging process, and tag appearance. The experimental results demonstrate that the new method improves the performance of even the best of the four previous methods. Yielding higher consistency, accuracy, and intrinsic tag reliability assessment, the proposed method allows for improved analysis of cardiac motion.     

64层CT冠状静脉系统成像在心脏电生理治疗中的应用价值探讨

目的评价64层CT显示心脏冠状静脉系统(CS)及其分支解剖的能力及其在心脏电生理治疗中的应用价值。方法回顾分析同期行64层CT扫描与选择性冠状静脉造影(CVG)的患者46例,观察CS及其分支的走行;以CVG为金标准,计数CS及其各分支的显影率;测量CS口、心中静脉(MCV)口和左室后静脉(PV)口的直径,测量冠状静脉各主要分支间的距离,分析MSCT与CVG测量值间的相关性。结果64层CT CS及其各主要分支的显示率为100%。所示的CS及其分支的走行与CVG一致;两种方法的冠状静脉直径测量值呈高度线性正相关,相关系数分别为0.89(CS),0.83(MCV),0.95(PV),P<0.001。结论64层CT可以无创评价CS及其分支的解剖,在心脏电生理治疗中有重要意义。     

Respiratory liver motion tracking during transcatheter procedures using guidewire detection

Transcatheter chemoembolization of liver tumors is performed under X-ray fluoroscopic image guidance. This is a difficult procedure because the vessels of the liver are constantly moving due to respiration and they are not visible in the X-ray image unless a contrast medium is injected. In order to help the interventional radiologist during the treatment, we propose to superimpose on to the fluoroscopic image a pre-acquired contrast-enhanced 2D or 3D image while accounting for liver motion. Our approach proposes to track the guidewire from frame to frame. Our proposed method can be split into two steps. First the guidewire is automatically detected; then the motion between two frames is estimated using a robust ICP (iterative closest point) algorithm. We have tested our method on simulated X-ray fluoroscopic images of a moving guidewire and applied it on 4 clinical sequences. Simulation demonstrated that the mean precision of our method is inferior to 1 mm. On clinical data, preliminary results demonstrated that this method allows for respiratory motion compensation of liver vessels with a mean accuracy inferior to 3 mm.     

非刚性配准运动校正算法与双扇区重组对改善冠状动脉CTA图像质量的对比分析

目的一种基于非刚性配准的运动校正算法智能边缘修复技术(intelligent boundary registration,IBR)已应用于冠状动脉CTA成像。通过与双扇区重组图像质量的比较,评估IBR技术应用于冠状动脉成像的效果。方法本回顾性研究经医院伦理委员会批准,并获得患者的知情同意。收集本院行能谱CT冠状动脉CTA检查且平均心率为65次/min(范围58~75次/min)的70例可疑心肌缺血患者,在最佳心动时相进行双扇区图像重组(SSB2),即为SSB2组;在最佳心动时相进行单扇区图像重组,基于最佳单扇区重组执行IBR重组以生成IBR图像,即为IBR组。采用5分制评分(5=图像质量优秀;1=不能满足诊断要求),评价指标包括:整体图像质量、血管的评估(连续性、有无运动伪影、血管的边缘模糊与否)。由两位心血管放射学医师通过双盲和独立观察比较,对SSB2和IBR处理后的图像分别进行冠状动脉整体水平和冠状动脉节段评分。结果共分析了70位患者冠状动脉的984个节段。两位医师的评分结果具有良好的一致性(k0.81),SSB2组和IBR组的平均评分分别为(4.02±1.28)和(4.45±1.01),两者间的差异具有统计学差异(Z=-9.22,P0.01)。在基于冠状动脉节段的分析中,可接受的图像质量不低于3分的百分比分别为88.3%和91.2%,具有统计学差异(X~2=5.68,P0.05);可接受的图像质量不低于4分的百分比为分别为73.3%和88.5%,具有统计学差异(X~2=6.47,P0.05)。在重组的血管中不可评估的节段IBR组明显低于SSB2组,差异具有统计学意义(4.2%与9.5%,X~2=12.13,P0.01)。结论能谱CT冠状动脉IBR技术可以提高冠状动脉CTA的图像质量,减少阶梯状伪影。     

Segmentation of lumen and outer wall of abdominal aortic aneurysms from 3D black-blood MRI with a registration based geodesic active contour model

Segmentation of the geometric morphology of abdominal aortic aneurysm is important for interventional planning. However, the segmentation of both the lumen and the outer wall of aneurysm in magnetic resonance (MR) image remains challenging. This study proposes a registration based segmentation methodology for efficiently segmenting MR images of abdominal aortic aneurysms. The proposed methodology first registers the contrast enhanced MR angiography (CE-MRA) and black-blood MR images, and then uses the Hough transform and geometric active contours to extract the vessel lumen by delineating the inner vessel wall directly from the CE-MRA. The proposed registration based geometric active contour is applied to black-blood MR images to generate the outer wall contour. The inner and outer vessel wall are then fused presenting the complete vessel lumen and wall segmentation. The results obtained from 19 cases showed that the proposed registration based geometric active contour model was efficient and comparable to manual segmentation and provided a high segmentation accuracy with an average Dice value reaching 89.79%.