Jingfan Fan

Local statistics and non-local mean filter for speckle noise reduction in medical ultrasound image

Medical ultrasound images are corrupted by speckle noise, which is multiplicative. This noise limits the contrast resolution in these images and complicates image-based quantitative measurement and diagnosis. In this study, the speckle noise in the ultrasound image is modeled by local statistics of the intensity distribution. And the non-local mean (NLM) filter is utilized to filter additional noise by applying the redundancy information in noisy images. A hybrid denoising method is proposed in consideration of the characteristics of both the local statistics of speckle noise and the NLM filter. The study combines local statistics with the NLM filter to reduce speckle in ultrasound images. The local statistics of speckle noise is estimated by local patches, while the intensity of the denoising pixel is computed by the weighted average of all the pixels by using the NLM. The weight is determined according to the similarity measures between the intensities of the local patches. The performance of the proposed method is evaluated on synthetic data, simulated images, and real images. Results of quantitative analysis and visual inspection of the synthetic data and of the real images demonstrate that the proposed method outperforms the original NLM, as well as many previously developed methods.

External force back-projective composition and globally deformable optimization for 3-D coronary artery reconstruction

The clinical value of the 3D reconstruction of a coronary artery is important for the diagnosis and intervention of cardiovascular diseases. This work proposes a method based on a deformable model for reconstructing coronary arteries from two monoplane angiographic images acquired from different angles. First, an external force back-projective composition model is developed to determine the external force, for which the force distributions in different views are back-projected to the 3D space and composited in the same coordinate system based on the perspective projection principle of x-ray imaging. The elasticity and bending forces are composited as an internal force to maintain the smoothness of the deformable curve. Second, the deformable curve evolves rapidly toward the true vascular centerlines in 3D space and angiographic images under the combination of internal and external forces. Third, densely matched correspondence among vessel centerlines is constructed using a curve alignment method. The bundle adjustment method is then utilized for the global optimization of the projection parameters and the 3D structures. The proposed method is validated on phantom data and routine angiographic images with consideration for space and re-projection image errors. Experimental results demonstrate the effectiveness and robustness of the proposed method for the reconstruction of coronary arteries from two monoplane angiographic images. The proposed method can achieve a mean space error of 0.564 mm and a mean re-projection error of 0.349 mm.

Brain MR image denoising for Rician noise using pre-smooth non-local means filter

BackgroundMagnetic resonance imaging (MRI) is corrupted by Rician noise, which is image dependent and computed from both real and imaginary images. Rician noise makes image-based quantitative measurement difficult. The non-local means (NLM) filter has been proven to be effective against additive noise.MethodsConsidering the characteristics of both Rician noise and the NLM filter, this study proposes a frame for a pre-smoothing NLM (PSNLM) filter combined with image transformation. In the PSNLM frame, noisy MRI is first transformed into an image in which noise can be treated as additive noise. Second, the transformed MRI is pre-smoothed via a traditional denoising method. Third, the NLM filter is applied to the transformed MRI, with weights that are computed from the pre-smoothed image. Finally, inverse transformation is performed on the denoised MRI to obtain the denoising results.ResultsTo test the performance of the proposed method, both simulated and real patient data are used, and various pre-smoothing (Gaussian, median, and anisotropic filters) and image transformation [squared magnitude of the MRI, and forward and inverse variance-stabilizing trans-formations (VST)] methods are used to reduce noise. The performance of the proposed method is evaluated through visual inspection and quantitative comparison of the peak signal-to-noise ratio of the simulated data. The real data include Alzheimer’s disease patients and normal controls. For the real patient data, the performance of the proposed method is evaluated by detecting atrophy regions in the hippocampus and the parahippocampal gyrus.ConclusionsThe comparison of the experimental results demonstrates that using a Gaussian pre-smoothing filter and VST produce the best results for the peak signal-to-noise ratio (PSNR) and atrophy detection.

Convex hull indexed Gaussian mixture model (CH-GMM) for 3D point set registration

To solve the problem of rigid/non-rigid 3D point set registration, a novel convex hull indexed Gaussian mixture model (CH-GMM) is proposed in this paper. The model works by computing a weighted Gaussian mixture model (GMM) response over the convex hull of each point set. Three conditions, proximity, area conservation and projection consistency, are incorporated into the model so as to improve its performance. Given that the convex hull is the tightest convex set of a point set, the combination of Gaussian mixture and convex hull can effectively preserve the topological structure of a point set. Furthermore, computational complexity can be significantly reduced since only the GMM of the convex hull (instead of the whole point set) needs to be calculated. Rigid registration is achieved by seeking the best rigid transformation parameters yielding the most similar CH-GMM responses. Non-rigid deformation is realized by optimizing the coordinates of the control points used by the thin-plate spline model for interpolating the entire point set. Experiments are designed to evaluate a methods robustness to rotational changes between two point sets, positional noise, differences in density and partial overlap. The results demonstrated better robustness and registration accuracy of CH-GMM based method over state-of-the-art methods including iterative closest point, coherent point drift and the GMM method. Besides, the computation of CH-GMM is efficient. A novel CH-GMM method is proposed for non-rigid registration of point sets.The CH-GMM can effectively present the topological information of point set.The CH-GMM matching of the point sets is very fast.The registration is very robust for points with large differences.

3-Points Convex Hull Matching (3PCHM) for fast and robust point set registration

Point set registration plays a crucial role in numerous computer vision applications. This paper proposes a novel and general approach called three-point convex hull matching (3PCHM) for registering two point sets with similarity transform. First, convex hulls are extracted from both point sets. Triangular patches on the surface of convex hulls are specified by predefining their normal vectors, thus guaranteeing that all points are located on the same side of any randomly selected triangle plane. Second, the potential similar triangle pair set is obtained by comparing the length ratio of the edges on the two extracted convex hulls. Thereafter, the transformation parameters for each pairwise triangle are calculated by minimizing the Euclidean distance between the corresponding vertex pairs. Furthermore, a k-dimensional (k-d) tree is used to accelerate the closest point search for the whole point sets. Third, outliers that may lead to significant errors are discarded by integrating the random sample consensus algorithm for global optimization. Experiments show that the proposed 3PCHM is robust even with the existence of noise and outliers and is effective in cases of part-to-part registration and part-to-whole registration.

Augmented reality based real-time subcutaneous vein imaging system.

A novel 3D reconstruction and fast imaging system for subcutaneous veins by augmented reality is presented. The study was performed to reduce the failure rate and time required in intravenous injection by providing augmented vein structures that back-project superimposed veins on the skin surface of the hand. Images of the subcutaneous vein are captured by two industrial cameras with extra reflective near-infrared lights. The veins are then segmented by a multiple-feature clustering method. Vein structures captured by the two cameras are matched and reconstructed based on the epipolar constraint and homographic property. The skin surface is reconstructed by active structured light with spatial encoding values and fusion displayed with the reconstructed vein. The vein and skin surface are both reconstructed in the 3D space. Results show that the structures can be precisely back-projected to the back of the hand for further augmented display and visualization. The overall system performance is evaluated in terms of vein segmentation, accuracy of vein matching, feature points distance error, duration times, accuracy of skin reconstruction, and augmented display. All experiments are validated with sets of real vein data. The imaging and augmented system produces good imaging and augmented reality results with high speed.

Fast multi-scale feature fusion for ECG heartbeat classification

Electrocardiogram (ECG) is conducted to monitor the electrical activity of the heart by presenting small amplitude and duration signals; as a result, hidden information present in ECG data is difficult to determine. However, this concealed information can be used to detect abnormalities. In our study, a fast feature-fusion method of ECG heartbeat classification based on multi-linear subspace learning is proposed. The method consists of four stages. First, baseline and high frequencies are removed to segment heartbeat. Second, as an extension of wavelets, wavelet-packet decomposition is conducted to extract features. With wavelet-packet decomposition, good time and frequency resolutions can be provided simultaneously. Third, decomposed confidences are arranged as a two-way tensor, in which feature fusion is directly implemented with generalized N dimensional ICA (GND-ICA). In this method, co-relationship among different data information is considered, and disadvantages of dimensionality are prevented; this method can also be used to reduce computing compared with linear subspace-learning methods (PCA). Finally, support vector machine (SVM) is considered as a classifier in heartbeat classification. In this study, ECG records are obtained from the MIT-BIT arrhythmia database. Four main heartbeat classes are used to examine the proposed algorithm. Based on the results of five measurements, sensitivity, positive predictivity, accuracy, average accuracy, and t-test, our conclusion is that a GND-ICA-based strategy can be used to provide enhanced ECG heartbeat classification. Furthermore, large redundant features are eliminated, and classification time is reduced.

Shape context and projection geometry constrained vasculature matching for 3D reconstruction of coronary artery

Vascular structure matching in X-ray angiographic images sequence obtained at different imaging angles is important for identifying vascular structures in computer-assisted diagnosis of coronary artery diseases. In this paper, a novel shape context and projection geometry constraint method is proposed for soft matching and identification of coronary artery structures. Firstly, the matching energy function between vasculatures in different angiographic views is constructed based on the local geometry constraint and perspective projection constraint. Secondly, initial matching of the vasculatures is established by the shape context constraint. Thirdly, the deterministic annealing method is used to optimize the matching function. Hence, optimal correspondences are obtained by iteratively reducing the temperature of the optimization function. Finally, on the basis of the obtained correspondences, 3D coronary artery structure can be reconstructed according to the theory of binocular stereo vision in computer vision. Experiments show that the relationship among the different views can be accurately constructed by the proposed method. The proposed method is fully automatic, so it can assist physician to rapidly identify and correlate vascular structures from angiographic images obtained at different imaging angles.

Adaptive Tensor-Based Principal Component Analysis for Low-Dose CT Image Denoising.

Computed tomography (CT) has a revolutionized diagnostic radiology but involves large radiation doses that directly impact image quality. In this paper, we propose adaptive tensor-based principal component analysis (AT-PCA) algorithm for low-dose CT image denoising. Pixels in the image are presented by their nearby neighbors, and are modeled as a patch. Adaptive searching windows are calculated to find similar patches as training groups for further processing. Tensor-based PCA is used to obtain transformation matrices, and coefficients are sequentially shrunk by the linear minimum mean square error. Reconstructed patches are obtained, and a denoised image is finally achieved by aggregating all of these patches. The experimental results of the standard test image show that the best results are obtained with two denoising rounds according to six quantitative measures. For the experiment on the clinical images, the proposed AT-PCA method can suppress the noise, enhance the edge, and improve the image quality more effectively than NLM and KSVD denoising methods.

Convex hull matching and hierarchical decomposition for multimodality medical image registration.

This study proposes a novel hierarchical pyramid strategy for 3D registration of multimodality medical images. The surfaces of the source and target volume data are first extracted, and the surface point clouds are then aligned roughly using convex hull matching. The convex hull matching registration procedure could align images with large-scale transformations. The original images are divided into blocks and the corresponding blocks in the two images are registered by affine and non-rigid registration procedures. The sub-blocks are iteratively smoothed by the Gaussian kernel with different sizes during the registration procedure. The registration result of the large kernel is taken as the input of the small kernel registration. The fine registration of the two volume data sets is achieved by iteratively increasing the number of blocks, in which increase in similarity measure is taken as a criterion for acceptation of each iteration level. Results demonstrate the effectiveness and robustness of the proposed method in registering the multiple modalities of medical images.