Graph Cut Method Research Articles

Glioma extraction from MR images employing Gradient Based Kernel Selection Graph Cut technique

Medical imaging is one of the most daunting, challenging, and emerging research topics in image processing. Segmenting the glioma from the brain magnetic resonance images (MRI) is an important and demanding task, as it assists the medical experts for the disease diagnosis process. Recent research methods in image segmentation have highlighted the prospective of graph-based techniques for medical applications. As graph cut (GC) method is interactive in nature, it requires manual selection of the initial kernels for processing. The popularity of the GC method is limited by the occurrence of small cuts due to its shrinkage behavior leading to inaccurate extraction causing erroneous regions. This paper addresses the open research issue of shrinkage behavior by proposing the gradient based kernel selection (GBKS) GC method emphasizing on the directive inclination of the intensity scales. The proposed technique aids in the initialization of GC, removes the shrinkage problem, and locates the tumor in brain images without any human intervention. The performance results of the proposed GBKS GC method are evaluated on high-grade glioma and low-grade glioma MRI images and are analyzed and compared by using various measures. All the results present a remarkable improvement with GBKS GC technique over other existing techniques.

The Influence of Different Segmentation Methods on the Extraction of Imaging Histological Features of Hepatocellular Carcinoma CT.

In order to analyze the influence of different segmentation techniques on hepatocellular carcinoma (HCC) CT (Computed Tomography) imaging histological feature extraction, Grow Cut method and Graph Cut method are used to segment hepatocellular carcinoma from arterial CT images of HCC patients, and the stability and repeatability of imaging histological features are studied. Meanwhile, hierarchical clustering method is used to reduce the redundancy of features. The results show that the repeatability and redundancy mainly depend on the method of tumor segmentation. Semi-automatic segmentation method can improve the repeatability of image features, and hierarchical clustering can reduce the redundancy of features. Different segmentation techniques have different effects on the extraction of histological features of CT images of HCC.

Evaluation of the performance of classification algorithms for XFEL single-particle imaging data.

Using X-ray free-electron lasers (XFELs), it is possible to determine three-dimensional structures of nanoscale particles using single-particle imaging methods. Classification algorithms are needed to sort out the single-particle diffraction patterns from the large amount of XFEL experimental data. However, different methods often yield inconsistent results. This study compared the performance of three classification algorithms: convolutional neural network, graph cut and diffusion map manifold embedding methods. The identified single-particle diffraction data of the PR772 virus particles were assembled in the three-dimensional Fourier space for real-space model reconstruction. The comparison showed that these three classification methods lead to different datasets and subsequently result in different electron density maps of the reconstructed models. Interestingly, the common dataset selected by these three methods improved the quality of the merged diffraction volume, as well as the resolutions of the reconstructed maps.

A U-Net Deep Learning Framework for High Performance Vessel Segmentation in Patients With Cerebrovascular Disease.

Brain vessel status is a promising biomarker for better prevention and treatment in cerebrovascular disease. However, classic rule-based vessel segmentation algorithms need to be hand-crafted and are insufficiently validated. A specialized deep learning method—the U-net—is a promising alternative. Using labeled data from 66 patients with cerebrovascular disease, the U-net framework was optimized and evaluated with three metrics: Dice coefficient, 95% Hausdorff distance (95HD) and average Hausdorff distance (AVD). The model performance was compared with the traditional segmentation method of graph-cuts. Training and reconstruction was performed using 2D patches. A full and a reduced architecture with less parameters were trained. We performed both quantitative and qualitative analyses. The U-net models yielded high performance for both the full and the reduced architecture: A Dice value of ~0.88, a 95HD of ~47 voxels and an AVD of ~0.4 voxels. The visual analysis revealed excellent performance in large vessels and sufficient performance in small vessels. Pathologies like cortical laminar necrosis and a rete mirabile led to limited segmentation performance in few patients. The U-net outperfomed the traditional graph-cuts method (Dice ~0.76, 95HD ~59, AVD ~1.97). Our work highly encourages the development of clinically applicable segmentation tools based on deep learning. Future works should focus on improved segmentation of small vessels and methodologies to deal with specific pathologies.

Parallelization of Jump Flood-Cut Method in Shared Memory Multicore Systems

Graph cut approach has evidenced to be trenchant strategy for the segmentation of digital images. Primary restraint of existing graph cut method is that of acute arithmetic calculation which use up tremendous processing time to carry into action. This paper aims on parallelizing jump flooding technique and push re-label scheme with NSP and BFS methodology on shared memory multicore system using OpenMP compiler directives. OpenMP API includes a rich set of library routines and environment variables which enables the parallelization of JF-Cut method within a framework where parallelism is achieved by sharing a single common memory among multiple co-processors. OpenMP remains live area of interest for research in segmentation of digital images as it accelerates computational speed of complex scientific applications like image segmentation using multicore architecture in shared memory mode.

A comparative performance of gray level image thresholding using normalized graph cut based standard S membership function

In this research paper, we use a normalized graph cut measure as a thresholding principle to separate an object from the background based on the standard S membership function. The implementation of the proposed algorithm known as fuzzy normalized graph cut method. This proposed algorithm compared with the fuzzy entropy method [25], Kittler [11], Rosin [21], Sauvola [23] and Wolf [33] method. Moreover, we examine that in most cases, our algorithm gives the lowest absolute error that improves the segmentation process of gray images. Finally, we change different parameter values in fuzzy normalized graph cut and the effect of the substitutes is studied. Also, we analyze the computational complexity of fuzzy weight matrix (fuzzification) results with a weight matrix (classical) results.

Water-fat separation and parameter mapping in cardiac MRI via deep learning with a convolutional neural network.

Water-fat separation is a postprocessing technique most commonly applied to multiple-gradient-echo magnetic resonance (MR) images to identify fat, provide images with fat suppression, and to measure fat tissue concentration. Recently, Numerous advancements have been reported. In contrast to early methods, the process of water-fat separation has become complicated due to multiparametric analytic models, optimization methods, and the absence of a unified framework for diverse source data. To determine the feasibility and performance of MRI water-fat separation and parametric mapping via deep learning (DL) with a range of inputs. Retrospective data usage. Ninety cardiac MR examinations from normal control, acute, subacute, and chronic myocardial infarction subjects were obtained, providing 1200 multiple gradient-echo acquisitions. 1.5 T/2D multiple gradient-echo pulse sequence ASSESSMENT: Ground-truth training and validation water-fat separation were obtained using a graph cut method with R2 *, off-resonance correction, and a multipeak fat spectrum. U-Net DL training with single and multiecho, complex, and magnitude inputs were compared using quantitative and three-observer subjective analysis. DL methods' image structural similarity, and quantitative proton density fat fraction (PDFF), R2 *, and off-resonance quantitative values were statistically compared with the GraphCut reference standard using Student's t-test and Pearson's correlation. Myocardial fat deposition in chronic myocardial infarction and intramyocardial hemorrhage in acute myocardial infarction were well visualized in the DL results. Predicted values for R2 *, off-resonance, water, and fat signal intensities were well correlated with a conventional model-based water fat separation (R2 ≥ 0.97, P < 0.001) with appropriate inputs. DL parameter maps had a 14% higher signal-to-noise ratio (P < 0.001) when compared with a conventional method. DL water-fat separation is feasible with a wide range of inputs, while R2 * and off-resonance mapping requires multiple echoes and complex images. With appropriate inputs, DL provides quantitative and subjective results comparable to conventional model-based methods. 1 Technical Efficacy Stage: 1 J. Magn. Reson. Imaging 2019;50:655-665.

On the safety effects of spatially aggregated traffic flow parameters at macro-levels

Traffic flow parameters have been found to significantly affect crash risk at micro-levels. If such effects do exist at macro-levels, at least two benefits could be expected: (1) the performance and estimates of planning-based crash models could be improved and (2) useful safety knowledge could be provided for regional traffic management. In this article, a flow-based spatial unit was developed by a graph-cut minimization method, based on which regional management strategies are often applied. The graph-cut method partitioned the central area of Kunshan, China, into multiple sub-regions (i.e. graph-cut unit), considering traffic density homogeneity. Bayesian Poisson lognormal models with conditional autoregressive priors were utilized to examine the safety effects of traffic flow parameters, based on the traditional planning-based units and the flow-based graph-cut units. According to the results, no significant traffic flow effect was found for the traffic analysis zone–based model. Traffic flow parameters resulted in a decreased model performance and potential endogeneity issues for the census tract–based model. However, traffic flow effects were found significant for the graph-cut-based model, with an improved model performance. In general, the safety effects of macro-level traffic flow need to be considered for flow-based units developed for regional management.

Increasing Accuracy of Optimal Surfaces Using Min-Marginal Energies.

Optimal surface methods are a class of graph cut methods posing surface estimation as an n-ary ordered labeling problem. They are used in medical imaging to find interacting and layered surfaces optimally and in low order polynomial time. Representing continuous surfaces with discrete sets of labels, however, leads to discretization errors and, if graph representations are made dense, excessive memory usage. Limiting memory usage and computation time of graph cut methods are important and graphs that locally adapt to the problem has been proposed as a solution. Min-marginal energies computed using dynamic graph cuts offer a way to estimate solution uncertainty and these uncertainties have been used to decide where graphs should be adapted. Adaptive graphs, however, introduce extra parameters, complexity, and heuristics. We propose a way to use min-marginal energies to estimate continuous solution labels that does not introduce extra parameters and show empirically on synthetic and medical imaging datasets that it leads to improved accuracy. The increase in accuracy was consistent and in many cases comparable with accuracy otherwise obtained with graphs up to eight times denser, but with proportionally less memory usage and improvements in computation time.

Retinal Layer Segmentation in Optical Coherence Tomography Images

The four major causes of blindness are age-related diseases, out of which three affects the retina. These diseases, i.e., glaucoma, diabetic retinopathy, and age-related macular degeneration, require life-long treatment and cause irreversible blindness. Conversely, early diagnosis has been shown to curtail or prevent blindness and visual impairments. A critical element of the clinical diagnosis is the analysis of individual retinal layer properties, as the manifestation of the dominant eye diseases has been shown to correlate with structural changes to the retinal layers. Regrettably, manual segmentation is dependent on the ophthalmologist's level of expertise, and currently becoming impractical due to advancement in imaging modalities. Inherently, much research on computer-aided diagnostic methods is conducted to aid in extracting useful layer information from these images, which were inaccessible without these techniques. However, speckle noise and intensity inhomogeneity remain a challenge with a detrimental effect on the performance of automated methods. In this paper, we propose a method comprising of fuzzy image processing techniques and graph-cut methods to robustly segment optical coherence tomography (OCT) into five (5) distinct layers. Notably, the method establishes a specific region of interest to suppress the interference of speckle noise, while Fuzzy C-means is utilized to build data terms for better integration into the continuous max-flow to handle inhomogeneity. The method is evaluated on 225 OCT B-scan images, and promising experimental results were achieved. The method will allow for early diagnosis of major eye diseases by providing the basic, yet critical layer information necessary for an effective eye examination.

Computer-aided automated discrimination of Alzheimer's disease and its clinical progression in magnetic resonance images using hybrid clustering and game theory-based classification strategies

Early detection and identification of morphological differences in the brain is crucial for the pre-surgical planning of Alzheimer's disease treatment. Magnetic resonance imaging (MRI) can detect Alzheimer's disease as well as its severity levels in patients. An automatic segmentation of the grey matter, white matter, cerebrospinal fluid and hippocampus is required to obtain accurate volume of various brain matters. In this study, an effective segmentation and classification techniques are proposed to accurately distinguish the progress of Alzheimer's disease, mild cognitive impairment and normal control subjects. A hybrid segmentation technique is formulated with K-means clustering and graph-cut methods to perform segmentation. The clustered regions are assigned labels according to their features for the classification analysis. They are further classified as normal cognitive impaired, stable mild cognitive impaired, progressive mild cognitive impaired or Alzheimer's disease using the game theory classifier. The proposed method achieves an accuracy of about 85.5 %.

A Volumetric Fusing Method for TLS and SFM Point Clouds

A terrestrial laser scanning (TLS) point cloud acquired from a given ground view is incomplete because of severe occlusion and self-occlusion. The models reconstructed by aligning the cross-source point clouds [TLS and structure-from-motion (SFM) point clouds] provide a more complete large-scale outdoor scene. However, because of differences in nonrigid deformation, stratified redundancy of alignment is inevitable and ubiquitous. Therefore, this paper presents a volumetric fusing method for cross-source three-dimensional reconstructions. To eliminate the stratification of aligned cross-source point clouds, we propose a graph-cuts method with boundary constraints for blending the two cross-source point clouds. Then, to reduce the gaps that exist in the blending results, we develop a progressive migration method combined with the local average direction of normal vectors to smooth the unconnected boundary. Finally, experimental results demonstrate the effectiveness of eliminating stratification with the proposed blending algorithm, and the progressive migration method achieves a smooth connection in the boundary of the blended point clouds.

Approximate ground states of the random-field Potts model from graph cuts.

While the ground-state problem for the random-field Ising model is polynomial, and can be solved using a number of well-known algorithms for maximum flow or graph cut, the analog random-field Potts model corresponds to a multiterminal flow problem that is known to be NP-hard. Hence an efficient exact algorithm is very unlikely to exist. As we show here, it is nevertheless possible to use an embedding of binary degrees of freedom into the Potts spins in combination with graph-cut methods to solve the corresponding ground-state problem approximately in polynomial time. We benchmark this heuristic algorithm using a set of quasiexact ground states found for small systems from long parallel tempering runs. For a not-too-large number q of Potts states, the method based on graph cuts finds the same solutions in a fraction of the time. We employ the new technique to analyze the breakup length of the random-field Potts model in two dimensions.

New l&lt;sub&gt;2,1&lt;/sub&gt;-Norm Relaxation of Multi-Way Graph Cut for Clustering

The clustering methods have absorbed even-increasing attention in machine learning and computer vision communities in recent years. Exploring manifold information in multi-way graph cut clustering, such as ratio cut clustering, has shown its promising performance. However, traditional multi-way ratio cut clustering method is NP-hard and thus the spectral solution may deviate from the optimal one. In this paper, we propose a new relaxed multi-way graph cut clustering method, where l2,1-norm distance instead of squared distance is utilized to preserve the solution having much more clearer cluster structures. Furthermore, the resulting solution is constrained with normalization to obtain more sparse representation, which can encourage the solution to contain more discrete values with many zeros. For the objective function, it is very difficult to optimize due to minimizing the ratio of two non-smooth items. To address this problem, we transform the objective function into a quadratic problem on the Stiefel manifold (QPSM), and introduce a novel yet efficient iterative algorithm to solve it. Experimental results on several benchmark datasets show that our method significantly outperforms several state-of-the-art clustering approaches.

Implementation of Image Segmentation with Graph Cut Method

Implementation of Image Segmentation with Graph Cut Method

A long baseline global stereo matching based upon short baseline estimation

In global stereo vision, balancing the matching efficiency and computing accuracy seems to be impossible because they contradict each other. In the case of a long baseline, this contradiction becomes more prominent. In order to solve this difficult problem, this paper proposes a novel idea to improve both the efficiency and accuracy in global stereo matching for a long baseline. In this way, the reference images located between the long baseline image pairs are firstly chosen to form the new image pairs with short baselines. The relationship between the disparities of pixels in the image pairs with different baselines is revealed by considering the quantized error so that the disparity search range under the long baseline can be reduced by guidance of the short baseline to gain matching efficiency. Then, the novel idea is integrated into the graph cuts (GCs) to form a multi-step GC algorithm based on the short baseline estimation, by which the disparity map under the long baseline can be calculated iteratively on the basis of the previous matching. Furthermore, the image information from the pixels that are non-occluded under the short baseline but are occluded for the long baseline can be employed to improve the matching accuracy. Although the time complexity of the proposed method depends on the locations of the chosen reference images, it is usually much lower for a long baseline stereo matching than when using the traditional GC algorithm. Finally, the validity of the proposed method is examined by experiments based on benchmark datasets. The results show that the proposed method is superior to the traditional GC method in terms of efficiency and accuracy, and thus it is suitable for long baseline stereo matching.

Deep convolutional neural network processing of aerial stereo imagery to monitor vulnerable zones near power lines

The monitoring of vegetation near high-voltage transmission power lines and poles is tedious. Blackouts present a huge challenge to power distribution companies and often occur due to tree growth in hilly and rural areas. There are numerous methods of monitoring hazardous overgrowth that are expensive and time-consuming. Accurate estimation of tree and vegetation heights near power poles can prevent the disruption of power transmission in vulnerable zones. This paper presents a cost-effective approach based on a convolutional neural network (CNN) algorithm to compute the height (depth maps) of objects proximal to power poles and transmission lines. The proposed CNN extracts and classifies features by employing convolutional pooling inputs to fully connected data layers that capture prominent features from stereo image patches. Unmanned aerial vehicle or satellite stereo image datasets can thus provide a feasible and cost-effective approach that identifies threat levels based on height and distance estimations of hazardous vegetation and other objects. Results were compared with extant disparity map estimation techniques, such as graph cut, dynamic programming, belief propagation, and area-based methods. The proposed method achieved an accuracy rate of 90%.

A Dynamic Graph Cuts Method with Integrated Multiple Feature Maps for Segmenting Kidneys in 2D Ultrasound Images

Automatic segmentation of kidneys in ultrasound (US) images remains a challenging task because of high speckle noise, low contrast, and large appearance variations of kidneys in US images. Because texture features may improve the US image segmentation performance, we propose a novel graph cuts method to segment kidney in US images by integrating image intensity information and texture feature maps. We develop a new graph cuts-based method to segment kidney US images by integrating original image intensity information and texture feature maps extracted using Gabor filters. To handle large appearance variation within kidney images and improve computational efficiency, we build a graph of image pixels close to kidney boundary instead of building a graph of the whole image. To make the kidney segmentation robust to weak boundaries, we adopt localized regional information to measure similarity between image pixels for computing edge weights to build the graph of image pixels. The localized graph is dynamically updated and the graph cuts-based segmentation iteratively progresses until convergence. Our method has been evaluated based on kidney US images of 85 subjects. The imaging data of 20 randomly selected subjects were used as training data to tune parameters of the image segmentation method, and the remaining data were used as testing data for validation. Experiment results demonstrated that the proposed method obtained promising segmentation results for bilateral kidneys (average Dice index = 0.9446, average mean distance = 2.2551, average specificity = 0.9971, average accuracy = 0.9919), better than other methods under comparison (P < .05, paired Wilcoxon rank sum tests). The proposed method achieved promising performance for segmenting kidneys in two-dimensional US images, better than segmentation methods built on any single channel of image information. This method will facilitate extraction of kidney characteristics that may predict important clinical outcomes such as progression of chronic kidney disease.

Segmentation and tracking of lung nodules via graph-cuts incorporating shape prior and motion from 4D CT.

We have developed a robust tool for performing volumetric and temporal analysis of nodules from respiratory gated four-dimensional (4D) CT. The method could prove useful in IMRT of lung cancer. We modified the conventional graph-cuts method by adding an adaptive shape prior as well as motion information within a signed distance function representation to permit more accurate and automated segmentation and tracking of lung nodules in 4D CT data. Active shape models (ASM) with signed distance function were used to capture the shape prior information, preventing unwanted surrounding tissues from becoming part of the segmented object. The optical flow method was used to estimate the local motion and to extend three-dimensional (3D) segmentation to 4D by warping a prior shape model through time. The algorithm has been applied to segmentation of well-circumscribed, vascularized, and juxtapleural lung nodules from respiratory gated CT data. In all cases, 4D segmentation and tracking for five phases of high-resolution CT data took approximately 10 min on a PC workstation with AMD Phenom II and 32 GB of memory. The method was trained based on 500 breath-held 3D CT data from the LIDC data base and was tested on 17 4D lung nodule CT datasets consisting of 85 volumetric frames. The validation tests resulted in an average Dice Similarity Coefficient (DSC) = 0.68 for all test data. An important by-product of the method is quantitative volume measurement from 4D CT from end-inspiration to end-expiration which will also have important diagnostic value. The algorithm performs robust segmentation of lung nodules from 4D CT data. Signed distance ASM provides the shape prior information which based on the iterative graph-cuts framework is adaptively refined to best fit the input data, preventing unwanted surrounding tissue from merging with the segmented object.

Random field Ising model in a uniform magnetic field: Ground states, pinned clusters and scaling laws

In this paper, we study the random field Ising model (RFIM) in an external magnetic field h . A computationally efficient graph-cut method is used to study ground state (GS) morphologies in this system for three different disorder types: Gaussian, uniform and bimodal. We obtain the critical properties of this system and find that they are independent of the disorder type. We also study GS morphologies via pinned-cluster distributions, which are scale-free at criticality. The spin-spin correlation functions (and structure factors) are characterized by a roughness exponent [Formula: see text]. The corresponding scaling function is universal for all disorder types and independent of h.