Graph Segmentation Research Articles

Acoustic-Phonetic Approaches for Improving Segment-Based Speech Recognition for Large Vocabulary Continuous Speech

Segment-based speech recognition has shown to be a competitive alternative to the state-of-the-art HMM-based techniques. Its accuracies rely heavily on the quality of the segment graph from which the recognizer searches for the most likely recognition hypotheses. In order to increase the inclusion rate of actual segments in the graph, it is important to recover possible missing segments generated by segment-based segmentation algorithm. An aspect of this research focuses on determining the missing segments due to missed detection of segment boundaries. The acoustic discontinuities, together with manner-distinctive features are utilized to recover the missing segments. Another aspect of improvement to our segment-based framework tackles the restriction of having limited amount of training speech data which prevents the usage of more complex covariance matrices for the acoustic models. Feature dimensional reduction in the form of the Principal Component Analysis (PCA) is applied to enable the training of full covariance matrices and it results in improved segment-based phoneme recognition. Furthermore, to benefit from the fact that segment-based approach allows the integration of phonetic knowledge, we incorporate the probability of each segment being one type of sound unit of a certain specific common manner of articulation into the scoring of the segment graphs. Our experiment shows that, with the proposed improvements, our segment-based framework approximately increases the phoneme recognition accuracy by approximately 25% of the one obtained from the baseline segment-based speech recognition.

Spectral identification of topological domains.

Topological domains have been proposed as the backbone of interphase chromosome structure. They are regions of high local contact frequency separated by sharp boundaries. Genes within a domain often have correlated transcription. In this paper, we present a computational efficient spectral algorithm to identify topological domains from chromosome conformation data (Hi-C data). We consider the genome as a weighted graph with vertices defined by loci on a chromosome and the edge weights given by interaction frequency between two loci. Laplacian-based graph segmentation is then applied iteratively to obtain the domains at the given compactness level. Comparison with algorithms in the literature shows the advantage of the proposed strategy. An efficient algorithm is presented to identify topological domains from the Hi-C matrix. The Matlab source code and illustrative examples are available at http://bionetworks.ccmb.med.umich.edu/ : indikar@med.umich.edu Supplementary data are available at Bioinformatics online.

Vascular endothelial growth factor A as predictive marker for mTOR inhibition in relapsing high-grade serous ovarian cancer.

BackgroundDevelopment of resistance against first line drug therapy including cisplatin and paclitaxel in high-grade serous ovarian cancer (HGSOC) presents a major challenge. Identifying drug candidates breaking resistance, ideally combined with predictive biomarkers allowing precision use are needed for prolonging progression free survival of ovarian cancer patients.Modeling of molecular processes driving drug resistance in tumor tissue further combined with mechanism of action of drugs provides a strategy for identification of candidate drugs and associated predictive biomarkers.ResultsConsolidation of transcriptomics profiles and biomedical literature mining results provides 1242 proteins linked with ovarian cancer drug resistance. Integrating this set on a protein interaction network followed by graph segmentation results in a molecular process model representation of drug resistant HGSOC embedding 409 proteins in 24 molecular processes. Utilizing independent transcriptomics profiles with follow-up data on progression free survival allows deriving molecular biomarker-based classifiers for predicting recurrence under first line therapy. Biomarkers of specific relevance are identified in a molecular process encapsulating TGF-beta, mTOR, Jak-STAT and Neurotrophin signaling. Mechanism of action molecular model representations of cisplatin and paclitaxel embed the very same signaling components, and specifically proteins afflicted with the activation status of the mTOR pathway become evident, including VEGFA. Analyzing mechanism of action interference of the mTOR inhibitor sirolimus shows specific impact on the drug resistance signature imposed by cisplatin and paclitaxel, further holding evidence for a synthetic lethal interaction to paclitaxel mechanism of action involving cyclin D1.ConclusionsStratifying drug resistant high grade serous ovarian cancer via VEGFA, and specifically treating with mTOR inhibitors in case of activation of the pathway may allow adding precision for overcoming resistance to first line therapy.

Multi-scale context for scene labeling via flexible segmentation graph

Using contextual information for scene labeling has gained substantial attention in the fields of image processing and computer vision. In this paper, a fusion model using flexible segmentation graph (FSG) is presented to explore multi-scale context for scene labeling problem. Given a family of segmentations, the representation of FSG is established based on the spatial relationship of these segmentations. In the scenario of FSG, the labeling inference process is formulated as a contextual fusion model, trained from the discriminative classifiers. Compared to previous approaches, which usually employ Conditional Random Fields (CRFs) or hierarchical models to explore contextual information, our FSG representation is flexible and efficient without hierarchical constraint, allowing us to capture a wide variety of visual context for the task of scene labeling. Our approach yields state-of-the-art results on the MSRC dataset (21 classes) and the LHI dataset (15 classes), and near-record results on the SIFT Flow dataset (33 classes) and PASCAL VOC segmentation dataset (20 classes), while producing a 320×240 scene labeling in less than a second. A remarkable fact is that our approach also outperforms recent CNN-based methods.

Incorporating Local and Global Context for Better Automated Analysis of Colorectal Cancer on Digital Pathology Slides

Phenotypic information derived from visual characteristics of colorectal cancer (CRC) is routinely used for diagnosis and recommendations for treatment. Previously published studies show that the ratio of tissue types within CRC is prognostic. Such studies generate large amounts of data, combining expert classifications with x-y coordinates, which has previously been used to train image analysis algorithms. This paper describes extensions to algorithms employed in previously published work, using pixel clustering as a pre-processing step before normalised cuts in order to reduce the size of the graph for unsupervised segmentation. Image segments are processed for features and given a candidate classification which is weighted by neighbouring segment classes. Global slide features are incorporated to mitigate inconsistencies in overall appearance caused by histological and biological differences. The proposed algorithm increases agreement with the ground truth from 75% to 79% on a dataset of 7,159 images across 157 digital slides.

Fuzzy Neural Network Classification of all True Vessels from Segmented Retinal Images

In general, the risk on recognizing the Cardio Vascular (CV) is made with the retinal blood vessel tracking. If the wrong blood vessel is identified, then it may lead to an incorrect clinical diagnosis. In the work already available, the segmented vascular structure is modeled in the form of a vessel segment graph and then the problem of identification of the vessels is formulated. But in this technique, the vessel segmentation is carried out with the Optic Disc (OD) not being eliminated. As the OD remains as the converging point for all the vessels present inside the retina, the OD boundary elimination results in few of the visible blood vessels cross right through it. Therefore, the OD boundary has to be eliminated prior to the segmentation of the vessels, since it may cause OD pixels misdetection that overlap with the blood vessels. In this work, an automated segmentation and vessel segmentation based elimination of OD boundary is introduced to prevent such kind of misdetections that improve the accuracy further in the case of vessel detection. The proposed technique has made use of modules like pre-processing, segmentation and feature extraction. This paper presents a preprocessing step applying Kalman filtering for the purpose of the noise removal and normalization for achieving image enhancement. Optical disc segmentation step employs Discrete Anisotropic Filter (DAF) with Particle Swarm Optimization (PSO). Thereafter some of the true vessels (CV severity) are identified from the segmented images by means of Fuzzy Neural Network (FNN). The results obtained from experiments demonstrate that this algorithm proposed has been proven to be a highly efficient technique for the classification of retinal blood vessels and the average accuracy attained is 97.2%.

Automated Segmentation of Knee MRI Using Hierarchical Classifiers and Just Enough Interaction Based Learning: Data from Osteoarthritis Initiative.

We present a fully automated learning-based approach for segmenting knee cartilage in presence of osteoarthritis (OA). The algorithm employs a hierarchical set of two random forest classifiers. The first is a neighborhood approximation forest, the output probability map of which is utilized as a feature set for the second random forest (RF) classifier. The output probabilities of the hierarchical approach are used as cost functions in a Layered Optimal Graph Segmentation of Multiple Objects and Surfaces (LOGISMOS). In this work, we highlight a novel post-processing interaction called just-enough interaction (JEI) which enables quick and accurate generation of a large set of training examples. Disjoint sets of 15 and 13 subjects were used for training and tested on another disjoint set of 53 knee datasets. All images were acquired using double echo steady state (DESS) MRI sequence and are from the osteoarthritis initiative (OAI) database. Segmentation performance using the learning-based cost function showed significant reduction in segmentation errors (p < 0.05) in comparison with conventional gradient-based cost functions.

Reeb graph based segmentation of articulated components of 3D digital objects

Segmentation of the articulated components of digital objects by a fast and efficient algorithm is presented in this paper. Given a 3D object in the form of a triangulated mesh, the approach involves representation of the mesh in a topological space and relating quotient spaces, containing the topologically invariant sections of the object, to weighted Reeb graphs along each of yz-, zx-, and xy-planes. It is followed by segmentation of the Reeb graphs where the concept of exponential averaging for dynamic thresholding ensures natural segmentation with a relatively high degree of accuracy. The segmented quotient spaces corresponding to the three Reeb graphs are subsequently related to each other and transformed topologically to report the segmented object in an appropriate topological space. The segmentation is carried out in a framework of 3D grid so as to exploit the topological relation between the triangulated object and its tight isothetic cover, and hence involves efficient computation and storage. The accuracy of segmentation at different grid resolutions, its robustness w.r.t. rotation, and pose-invariance for a reasonable range of postures are demonstrated by experimental results on a variety of objects.

Topology polymorphism graph for lung tumor segmentation in PET-CT images

Accurate lung tumor segmentation is problematic when the tumor boundary or edge, which reflects the advancing edge of the tumor, is difficult to discern on chest CT or PET. We propose a ‘topo-poly’ graph model to improve identification of the tumor extent. Our model incorporates an intensity graph and a topology graph. The intensity graph provides the joint PET-CT foreground similarity to differentiate the tumor from surrounding tissues. The topology graph is defined on the basis of contour tree to reflect the inclusion and exclusion relationship of regions. By taking into account different topology relations, the edges in our model exhibit topological polymorphism. These polymorphic edges in turn affect the energy cost when crossing different topology regions under a random walk framework, and hence contribute to appropriate tumor delineation. We validated our method on 40 patients with non-small cell lung cancer where the tumors were manually delineated by a clinical expert. The studies were separated into an ‘isolated’ group (n = 20) where the lung tumor was located in the lung parenchyma and away from associated structures / tissues in the thorax and a ‘complex’ group (n = 20) where the tumor abutted / involved a variety of adjacent structures and had heterogeneous FDG uptake. The methods were validated using Dice’s similarity coefficient (DSC) to measure the spatial volume overlap and Hausdorff distance (HD) to compare shape similarity calculated as the maximum surface distance between the segmentation results and the manual delineations. Our method achieved an average DSC of 0.881 ± 0.046 and HD of 5.311 ± 3.022 mm for the isolated cases and DSC of 0.870 ± 0.038 and HD of 9.370 ± 3.169 mm for the complex cases. Student’s t-test showed that our model outperformed the other methods (p-values <0.05).

A global/local affinity graph for image segmentation.

Construction of a reliable graph capturing perceptual grouping cues of an image is fundamental for graph-cut based image segmentation methods. In this paper, we propose a novel sparse global/local affinity graph over superpixels of an input image to capture both short- and long-range grouping cues, and thereby enabling perceptual grouping laws, including proximity, similarity, continuity, and to enter in action through a suitable graph-cut algorithm. Moreover, we also evaluate three major visual features, namely, color, texture, and shape, for their effectiveness in perceptual segmentation and propose a simple graph fusion scheme to implement some recent findings from psychophysics, which suggest combining these visual features with different emphases for perceptual grouping. In particular, an input image is first oversegmented into superpixels at different scales. We postulate a gravitation law based on empirical observations and divide superpixels adaptively into small-, medium-, and large-sized sets. Global grouping is achieved using medium-sized superpixels through a sparse representation of superpixels' features by solving a ℓ0-minimization problem, and thereby enabling continuity or propagation of local smoothness over long-range connections. Small- and large-sized superpixels are then used to achieve local smoothness through an adjacent graph in a given feature space, and thus implementing perceptual laws, for example, similarity and proximity. Finally, a bipartite graph is also introduced to enable propagation of grouping cues between superpixels of different scales. Extensive experiments are carried out on the Berkeley segmentation database in comparison with several state-of-the-art graph constructions. The results show the effectiveness of the proposed approach, which outperforms state-of-the-art graphs using four different objective criteria, namely, the probabilistic rand index, the variation of information, the global consistency error, and the boundary displacement error.

Low-Level Hierarchical Multiscale Segmentation Statistics of Natural Images.

This paper is aimed at obtaining the statistics as a probabilistic model pertaining to the geometric, topological and photometric structure of natural images. The image structure is represented by its segmentation graph derived from the low-level hierarchical multiscale image segmentation. We first estimate the statistics of a number of segmentation graph properties from a large number of images. Our estimates confirm some findings reported in the past work, as well as provide some new ones. We then obtain a Markov random field based model of the segmentation graph which subsumes the observed statistics. To demonstrate the value of the model and the statistics, we show how its use as a prior impacts three applications: image classification, semantic image segmentation and object detection.

A New Framework for Direct Saliency Detection and Segmentation Based on Graph Methods

Saliency detection is an important research topic in computer vision. The traditional methods compute image saliency map, then salient segmentation is based on the corresponding saliency map. Unfortunately, overall performance of this method is poor due to the reason of losing some fine details and spatial information within image. This paper presents a new framework to overcome the drawback, named FDSRDS(Framework for Directly Salient Region Detection and Segmentation based on graph methods). Under FDSRSD, firstly, we get the foreground image by segmenting the original image via our extended grabcut algorithm. Mostly, the saliency region is within the foreground part. Secondly, we segment the foreground image into regions by means of graph based segmentation and nearest neighbor graph . Thirdly, we use relative weber's luminance rules to calculate every region’s luminance. Finally, we get the maximum luminance region which is the saliency region. Under FDSRSD framework, algorithms we proposed capture fine details and spatial relationships in saliency computation. We demonstrate impressive results by evaluating our method with other five state-of-the-art methods on the publicly available data set.

Genus zero graph segmentation: Estimation of intracranial volume

The intracranial volume (ICV) in children with premature fusion of one or more sutures in the calvaria is of interest due to the risk of increased intracranial pressure. Challenges for automatic estimation of ICV include holes in the skull e.g. the foramen magnum and fontanelles. In this paper, we present a fully automatic 3D graph-based method for segmentation of the ICV in non-contrast CT scans. We reformulate the ICV segmentation problem as an optimal genus 0 segmentation problem in a volumetric graph. The graph is the result of a volumetric spherical subsampling. The equidistantly sampled data points are connected using Delaunay tetrahedralisation creating a highly connected neighborhood. A Markov Random Field (MRF) is constructed on the graph with probabilities learned from an Expectation Maximisation algorithm matching a Mixture of Gaussians to the data. The result of the MRF segmentation is compared to manual segmentations performed by an expert. We have achieved very high Dice scores ranging from 98.14% to 99.00%, while volume deviation from the manual segmentation ranges from 0.7% to 3.7%. The Hausdorff distance, which shows the maximum error from automatic to manual segmentation, ranges from 4.73 to 9.81mm. Since this is sensitive to single error, we have also found the 95% Hausdorff distance, which ranges from 1.10 to 3.65mm. The segmentation is very consistent with the reference and differs only in difficult areas, where it seems that our method is much more slice-wise consistent than a manual segmentation. The proposed method is expected to perform well for other volumetric segmentations.

Automated Method for Identification and Artery-Venous Classification of Vessel Trees in Retinal Vessel Networks

The separation of the retinal vessel network into distinct arterial and venous vessel trees is of high interest. We propose an automated method for identification and separation of retinal vessel trees in a retinal color image by converting a vessel segmentation image into a vessel segment map and identifying the individual vessel trees by graph search. Orientation, width, and intensity of each vessel segment are utilized to find the optimal graph of vessel segments. The separated vessel trees are labeled as primary vessel or branches. We utilize the separated vessel trees for arterial-venous (AV) classification, based on the color properties of the vessels in each tree graph. We applied our approach to a dataset of 50 fundus images from 50 subjects. The proposed method resulted in an accuracy of 91.44 correctly classified vessel pixels as either artery or vein. The accuracy of correctly classified major vessel segments was 96.42.

LOGISMOS-B: layered optimal graph image segmentation of multiple objects and surfaces for the brain.

Automated reconstruction of the cortical surface is one of the most challenging problems in the analysis of human brain magnetic resonance imaging (MRI). A desirable segmentation must be both spatially and topologically accurate, as well as robust and computationally efficient. We propose a novel algorithm, LOGISMOS-B, based on probabilistic tissue classification, generalized gradient vector flows and the LOGISMOS graph segmentation framework. Quantitative results on MRI datasets from both healthy subjects and multiple sclerosis patients using a total of 16,800 manually placed landmarks illustrate the excellent performance of our algorithm with respect to spatial accuracy. Remarkably, the average signed error was only 0.084 mm for the white matter and 0.008 mm for the gray matter, even in the presence of multiple sclerosis lesions. Statistical comparison shows that LOGISMOS-B produces a significantly more accurate cortical reconstruction than FreeSurfer, the current state-of-the-art approach (p << 0.001). Furthermore, LOGISMOS-B enjoys a run time that is less than a third of that of FreeSurfer, which is both substantial, considering the latter takes 10 h/subject on average, and a statistically significant speedup.

Retinal Vessels Segmentation Using Supervised Classifiers for Identification of Cardio Vascular Diseases

The risk of cardio vascular diseases can be identified by measuring the retinal blood vessel. The identification of wrong blood vessel may result in wrong clinical diagnosis. This proposed system addresses the problem of identifying the true vessel by vascular structure segmentation. In this proposed model the segmented vascular structure is modelled as a vessel segment graph and the true vessels are identified by using supervised classifier approach. This paper proposes a post processing step in diagnose cardiovascular diseases which can be identified by tracking a true vessel from the optimal forest in the graph given a set on constraints.

Robust cortical thickness measurement with LOGISMOS-B.

Cortical thickness (CT) is an important morphometric measure that has implications for psychiatric and neurologic processes. We propose a novel approach for automatically computing CT in an accurate and robust manner using LOGISMOS-B: Layered Optimal Graph Image Segmentation of Multiple Objects and Surfaces for the Brain. LOGISMOS-B is a cortical surface segmentation method based on LOGISMOS graph segmentation and generalized gradient vector flows. We evaluate our method on two different datasets (n = 83 total). The results show that LOGISMOS-B is more accurate than the popular FreeSurfer (FS) method and provides more reliable thickness measurements across a variety of challenging images. LOGISMOS-B accurately recovers known CT patterns, both across cortical lobes and locally, such as between the banks of the central sulcus, in healthy subjects and MS patients. Manual landmarks indicate a signed surface distance of 0.081 +/- 0.447mm for WM and 0.018 +/- 0.498mm for LOGISMOS-B, compared to 0.263 +/- 0.452mm for WM and -0.167 +/- 0.556mm for GM for FS, highlighting the surface placement accuracy of LOGISMOS-B. Finally, a regresion study shows that LOGISMOS-B provides strong correlation with age and plausible annual thinning rates across the cortex, with locally discerning thinning patterns, in agreement with the literature.

Paired Regions for Shadow Detection and Removal

In this paper, we address the problem of shadow detection and removal from single images of natural scenes. Differently from traditional methods that explore pixel or edge information, we employ a region-based approach. In addition to considering individual regions separately, we predict relative illumination conditions between segmented regions from their appearances and perform pairwise classification based on such information. Classification results are used to build a graph of segments, and graph-cut is used to solve the labeling of shadow and nonshadow regions. Detection results are later refined by image matting, and the shadow-free image is recovered by relighting each pixel based on our lighting model. We evaluate our method on the shadow detection dataset in Zhu et al. In addition, we created a new dataset with shadow-free ground truth images, which provides a quantitative basis for evaluating shadow removal. We study the effectiveness of features for both unary and pairwise classification.

Hybrid Method for Retinal Image Segmentation and Identifying True Vessels

As digital imaging and computing power increasingly develop, so too does the potential to use these technologies in ophthalmology. Image processing, analysis and computer vision techniques are increasing in prominence in all fields of medical science, and are especially pertinent to modern ophthalmology, as it is heavily dependent on visually oriented signs. We describe a novel technique that utilizes the global information of the segmented vascular structure to correctly identify true vessels in a retinal image. The model segmented vascular structure as a vessel segment graph and transform the problem of identifying true vessels to that of finding an optimal forest in the graph. An objective function to score forests is designed based on directional information. Our proposed solution employs candidate generation and expert knowledge to prune the search space. Each vessel is tracked individually by repeatedly finding the next vessel point with a scoring function that considers the pixel intensity and orientation in the vicinity of the current point in the image. Bifurcations and crossovers are detected using some intensity profile. Tracking for the same vessel then continues along the most likely path. The importance of our proposed work disambiguate between vessels at bifurcations and crossovers, we need to figure out if linking a vessel segment to one vessel will lead to an adjacent vessel being wrongly identified .

RATS: Rapid Automatic Tissue Segmentation in rodent brain MRI

BackgroundHigh-field MRI is a popular technique for the study of rodent brains. These datasets, while similar to human brain MRI in many aspects, present unique image processing challenges. We address a very common preprocessing step, skull-stripping, which refers to the segmentation of the brain tissue from the image for further processing. While several methods exist for addressing this problem, they are computationally expensive and often require interactive post-processing by an expert to clean up poorly segmented areas. This further increases total processing time per subject. New methodWe propose a novel algorithm, based on grayscale mathematical morphology and LOGISMOS-based graph segmentation, which is rapid, robust and highly accurate. ResultsComparative results obtained on two challenging in vivo datasets, consisting of 22 T1-weighted rat brain images and 10 T2-weighted mouse brain images illustrate the robustness and excellent performance of the proposed algorithm, in a fraction of the computational time needed by existing algorithms. Comparison with existing methodsIn comparison to current state-of-the-art methods, our approach achieved average Dice similarity coefficient of 0.92±0.02 and average Hausdorff distance of 13.6±5.2 voxels (vs. 0.85±0.20, p<0.05 and 42.6±22.9, p≪0.001) for the rat dataset, and 0.96±0.01 and average Hausdorff distance of 21.6±12.7 voxels (vs. 0.93±0.01, p≪0.001 and 33.7±3.5, p≪0.001) for the mouse dataset. The proposed algorithm took approximately 90s per subject, compared to 10–20min for the neural-network based method and 30–90min for the atlas-based method. ConclusionsRATS is a robust and computationally efficient method for accurate rodent brain skull-stripping even in challenging data.