Showing papers by "Qi Zhang published in 2012"
01 Jan 2012
TL;DR: Results demonstrate that the MsFCM-PSO can provide an accurate tool for ultrasound image segmentation and outperformed the traditional fuzzy c-means methods by 39.6% and 13.6%, in terms of the Pratt's figure of merit and segmentation accuracy, respectively.
Abstract: A multi-scale fuzzy c-means method integrated with particle swarm optimization (MsFCM-PSO) is proposed for ultrasound image segmentation. First, speckle reducing anisotropic diffusion is used to suppress noise in an ultrasound image and construct a series of images at multiple scales. Then the particle swarm optimization is incorporated into the multi-scale fuzzy c-means (MsFCM) to search for the global optima of cluster centers and update the membership of each pixel in a coarse-to-fine fashion. Finally, the image is segmented by assigning each pixel to the cluster with the highest membership. The method was validated on both synthetic and in vivo ultrasound images. It outperformed the traditional fuzzy c-means methods including MsFCM by 39.6% and 13.6%, in terms of the Pratt's figure of merit and segmentation accuracy, respectively. These results demonstrate that the MsFCM-PSO can provide an accurate tool for ultrasound image segmentation.
15 citations
16 Oct 2012
TL;DR: In this paper, the edge detector in the traditional an isotropic diffusion was replaced by the McIlhagga edge detector to suppress the speckle noise in ultrasound images.
Abstract: A new speckle reduction method for ultrasound images is proposed based on the McIlhagga edge detector. The edge detector in the traditional an isotropic diffusion was replaced by the McIlhagga edge detector to suppress the speckle noise in ultrasound images. The numerical solution of the McIlhagga edge detector-based an isotropic diffusion (MAD) is derived. Both synthetic and real ultrasound images are used to evaluate the MAD method. The performance of the MAD is compared with six traditional image denoising methods. It is shown that the MAD method is superior to the traditional methods in both noise reduction and detail preservation.
5 citations
19 May 2012
TL;DR: Wang et al. as discussed by the authors used spatial correlation of time intensity curves to capture detailed information from spatial-temporal neighborhoods of plaques and detect initial contours of the plaques, and then the speckle reducing anisotropic diffusion is adopted to modify the edge map of the gradient vector flow snake, and the initial contour are deformed using the modified snake and converged to refined contours.
Abstract: It is valuable to segment carotid atherosclerotic plaques in contrast-enhanced ultrasound (CEUS) images. Traditional methods for plaque segmentation neglect the temporal information in video images, and it limits the accuracy of the segmentation. We propose an algorithm for accurate segmentation of plaques in CEUS images using spatial-temporal analysis and snakes. First, the spatial correlation of time intensity curves is used to capture detailed information from spatial-temporal neighborhoods of plaques and detect initial contours of the plaques. Then the speckle reducing anisotropic diffusion is adopted to modify the edge map of the gradient vector flow snake, and the initial contours are deformed using the modified snake and converged to refined contours. The proposed method was evaluated via 21 in vivo images, and it outperformed the boundary vector field snake by 0.06 mm, 2.0%, 0.04 mm and 5.3%, in terms of the mean distance error, relative mean distance error, mean signed distance error, and relative difference degree, respectively. The results revealed that the proposed method can accurately detect plaques and delineate their contours.
5 citations
01 Oct 2012
TL;DR: This study proposed to improve the SNR of contrast image based on anisotropic diffusion which de-noises the contrast image with local gradient information and simulation and animal studies were presented to evaluate theSNR improvement.
Abstract: As a two-dimensional, high-resolution, minimal-invasive monitoring method, laser speckle contrast imaging (LSCI) is widely used to measure the relative speed of cerebral blood flow (CBF) under different physiological and pathological states, i.e. before and after acupuncture. During the last decade, many improved methods (spatial, temporal, and several hybrid methods) for better estimating the contrast values have been proposed based on the updated theoretical explanation of speckle phenomenon. Temporal LSCI method is used to obtain high spatial resolution contrast image with limited temporal resolution. However, the temporal contrast image still contains much noise. There were very few studies of post-processing methods to further improve the SNR of contrast image. Based on the random property of speckle, characteristics of noise in tissue area and vessel areas are different and thus should be treated differently in smoothing and/or filtering when improving the SNR. In this study, we proposed to improve the SNR of contrast image based on anisotropic diffusion which de-noises the contrast image with local gradient information. Simulation and animal studies were presented to evaluate the SNR improvement of contrast image.
1 citations