Segmenting high-frequency intracardiac ultrasound images of myocardium into infarcted, ischemic, and normal regions

A fast technique for hyper-echoic region separation from brain ultrasound images using patch based thresholding and cubic B-spline based contour smoothing

Ultrasound image guided brain surgery (UGBS) requires an automatic and fast image segmentation method. The level-set and active contour based algorithms have been found to be useful for obtaining topology-independent boundaries between different image regions. But slow convergence limits their use in online US image segmentation. The performance of these algorithms deteriorates on US images because of the intensity inhomogeneity. This paper proposes an effective region-driven method for the segmentation of hyper-echoic (HE) regions suppressing the hypo-echoic and anechoic regions in brain US images. An automatic threshold estimation scheme is developed with a modified Niblack's approach. The separation of the hyper-echoic and non-hyper-echoic (NHE) regions is performed by successively applying patch based intensity thresholding and boundary smoothing. First, a patch based segmentation is performed, which separates roughly the two regions. The patch based approach in this process reduces the effect of intensity heterogeneity within an HE region. An iterative boundary correction step with reducing patch size improves further the regional topology and refines the boundary regions. For avoiding the slope and curvature discontinuities and obtaining distinct boundaries between HE and NHE regions, a cubic B-spline model of curve smoothing is applied. The proposed method is 50-100 times faster than the other level-set based image segmentation algorithms. The segmentation performance and the convergence speed of the proposed method are compared with four other competing level-set based algorithms. The computational results show that the proposed segmentation approach outperforms other level-set based techniques both subjectively and objectively.

An Automatic Localization Algorithm for Ultrasound Breast Tumors Based on Human Visual Mechanism

Human visual mechanisms (HVMs) can quickly localize the most salient object in natural images, but it is ineffective at localizing tumors in ultrasound breast images. In this paper, we research the characteristics of tumors, develop a classic HVM and propose a novel auto-localization method. Comparing to surrounding areas, tumors have higher global and local contrast. In this method, intensity, blackness ratio and superpixel contrast features are combined to compute a saliency map, in which a Winner Take All algorithm is used to localize the most salient region, which is represented by a circle. The results show that the proposed method can successfully avoid the interference caused by background areas of low echo and high intensity. The method has been tested on 400 ultrasound breast images, among which 376 images succeed in localization. This means this method has a high accuracy of 94.00%, indicating its good performance in real-life applications.

Classification and segmentation of intracardiac masses in cardiac tumor echocardiograms

This paper describes an automatic method for classification and segmentation of different intracardiac masses in tumor echocardiograms. Identification of mass type is highly desirable, since to different treatment options for cardiac tumors (surgical resection) and thrombi (effective anticoagulant treatment) are possible. Correct diagnosis of the character of intracardiac mass in a living patient is a true challenge for a cardiologist; therefore, an objective image analysis method may be useful in heart diseases diagnosis. Image texture analysis is used to distinguish various types of masses. The presented methods assume that image texture encodes important histological features of masses and, therefore, texture numerical parameters enable the discrimination and segmentation of a mass. The recently developed technique based on the network of synchronized oscillators is proposed for the image segmentation. This technique is based on a 'temporary correlation' theory, which attempts to explain scene recognition as it would be performed by a human brain. This theory assumes that different groups of neural cells encode different properties of homogeneous image regions (e.g. shape, color, texture). Monitoring of temporal activity of cell groups leads to scene segmentation. A network of synchronized oscillators was successfully used for segmentation of Brodatz textures and medical textured images. The advantage of this network is its ability to detect texture boundaries. It can be also manufactured as a VLSI chip, for a very fast image segmentation. The accuracy of locating of analyzed tissues in the image should be assessed to evaluate a segmentation technique. The new evaluation method based on measurement of physical textured test objects was proposed. Firstly, a series of object images was obtained by the use of different devices (scanner, digital camera and TV camera). Secondly, the images were segmented using oscillator network and feedforward artificial neural network. Thirdly, geometrical test object parameters were estimated and compared to its true values. The experiment was repeated also for ultrasound images, which represented rectangular cross-section of synthetic sponge submerged in water. In addition, classification and segmentation of selected benign tumor echocardiograms were performed. Oscillator network was used with network weights defined for both whole texture region and texture boundary detection for the tumor segmentation. The latter method provides much faster segmentation with the similar accuracy. The obtained segmentation results were discussed and compared to the artificial neural network classifier. Finally, it was demonstrated that the network of synchronized oscillators is a reliable tool for the segmentation of the selected intracardiac masses, since it gives a relatively accurate location of analyzed tissues. The advantage of the proposed method is its resistance to changes of the visual information in the analyzed image and to noise and artifacts, often present in echocardiograms.

An efficient parameterless quadrilateral-based image segmentation method

This paper proposes a general quadrilateral-based framework for image segmentation, in which quadrilaterals are first constructed from an edge map, where neighboring quadrilaterals with similar features of interest are then merged together to form regions. Under the proposed framework, the quadrilaterals enable the elimination of local variations and unnecessary details for merging from which each segmented region is accurately and completely described by a set of quadrilaterals. To illustrate the effectiveness of the proposed framework, we derived an efficient and high-performance parameterless quadrilateral-based segmentation algorithm from the framework. The proposed algorithm shows that the regions obtained under the framework are segmented into multiple levels of quadrilaterals that accurately represent the regions without severely over or undersegmenting them. When evaluated objectively and subjectively, the proposed algorithm performs better than three other segmentation techniques, namely, seeded region growing, K-means clustering and constrained gravitational clustering, and offers an efficient description of the segmented objects conducive to content-based applications.

Monitoring structural changes in cells with high-frequency ultrasound signal statistics

We investigate the use of signal envelope statistics to monitor and quantify structural changes during cell death using an in vitro cell model. Using a f/2.35 transducer (center frequency 20 MHz), ultrasound backscatter data were obtained from pellets of acute myeloid leukemia cells treated with a DNA-intercolating chemotherapy drug, as well as from pellets formed with mixtures of treated and untreated cells. Simulations of signals from pellets of mixtures of cells were generated as a summation of point scatterers. The signal envelope statistics were examined by fitting the Rayleigh and generalized gamma distributions. The fit parameters of the generalized gamma distribution showed sensitivity to structural changes in the cells. The scale parameter showed a 200% increase (p<0.05) between untreated and cells treated for 24 h. The shape parameter showed a 50% increase (p<0.05) over 24 h. Experimental results showed reasonable agreement with simulations. The results indicate that high-frequency ultrasound signal statistics can be used to monitor structural changes within a very low percentage of treated cells in a population, raising the possibility of using this technique in vivo.

Combinative multi-scale level set framework for echocardiographic image segmentation

In the automatic segmentation of echocardiographic images, a priori shape knowledge has been used to compensate for poor features in ultrasound images. This shape knowledge is often learned via an off-line training process, which requires tedious human effort and is highly expertise-dependent. More importantly, a learned shape template can only be used to segment a specific class of images with similar boundary shape. In this paper, we present a multi-scale level set framework for segmentation of endocardial boundaries at each frame in a multiframe echocardiographic image sequence. We point out that the intensity distribution of an ultrasound image at a very coarse scale can be approximately modeled by Gaussian. Then we combine region homogeneity and edge features in a level set approach to extract boundaries automatically at this coarse scale. At finer scale levels, these coarse boundaries are used to both initialize boundary detection and serve as an external constraint to guide contour evolution. This constraint functions similar to a traditional shape prior. Experimental results validate this combinative framework.

Combining low-, high-level and empirical domain knowledge for automated segmentation of ultrasonic breast lesions

Breast cancer is the most frequently diagnosed malignancy and the second leading cause of mortality in women. In the last decade, ultrasound along with digital mammography has come to be regarded as the gold standard for breast cancer diagnosis. Automatically detecting tumors and extracting lesion boundaries in ultrasound images is difficult due to their specular nature and the variance in shape and appearance of sonographic lesions. Past work on automated ultrasonic breast lesion segmentation has not addressed important issues such as shadowing artifacts or dealing with similar tumor like structures in the sonogram. Algorithms that claim to automatically classify ultrasonic breast lesions, rely on manual delineation of the tumor boundaries. In this paper, we present a novel technique to automatically find lesion margins in ultrasound images, by combining intensity and texture with empirical domain specific knowledge along with directional gradient and a deformable shape-based model. The images are first filtered to remove speckle noise and then contrast enhanced to emphasize the tumor regions. For the first time, a mathematical formulation of the empirical rules used by radiologists in detecting ultrasonic breast lesions, popularly known as the "Stavros Criteria" is presented in this paper. We have applied this formulation to automatically determine a seed point within the image. Probabilistic classification of image pixels based on intensity and texture is followed by region growing using the automatically determined seed point to obtain an initial segmentation of the lesion. Boundary points are found on the directional gradient of the image. Outliers are removed by a process of recursive refinement. These boundary points are then supplied as an initial estimate to a deformable model. Incorporating empirical domain specific knowledge along with low and high-level knowledge makes it possible to avoid shadowing artifacts and lowers the chance of confusing similar tumor like structures for the lesion. The system was validated on a database of breast sonograms for 42 patients. The average mean boundary error between manual and automated segmentation was 6.6 pixels and the normalized true positive area overlap was 75.1%. The algorithm was found to be robust to 1) variations in system parameters, 2) number of training samples used, and 3) the position of the seed point within the tumor. Running time for segmenting a single sonogram was 18 s on a 1.8-GHz Pentium machine.