Shape-based Method Research Articles

Application of Support Vector Machine to Three-Dimensional Shape-Based Virtual Screening Using Comprehensive Three-Dimensional Molecular Shape Overlay with Known Inhibitors

In this study, machine learning using support vector machine was combined with three-dimensional (3D) molecular shape overlay, to improve the screening efficiency. Since the 3D molecular shape overlay does not use fingerprints or descriptors to compare two compounds, unlike 2D similarity methods, the application of machine learning to a 3D shape-based method has not been extensively investigated. The 3D similarity profile of a compound is defined as the array of 3D shape similarities with multiple known active compounds of the target protein and is used as the explanatory variable of support vector machine. As the measures of 3D shape similarity for our new prediction models, the prediction performances of the 3D shape similarity metrics implemented in ROCS, such as ShapeTanimoto and ScaledColor, were validated, using the known inhibitors of 15 target proteins derived from the ChEMBL database. The learning models based on the 3D similarity profiles stably outperformed the original ROCS when more than 10 known inhibitors were available as the queries. The results demonstrated the advantages of combining machine learning with the 3D similarity profile to process the 3D shape information of plural active compounds.

A Shape-Based Two-Stage Product Image Retrieval Method

Product image retrieval using content of the image is valuable for E-commerce application. But both search efficiency and accuracy are challenging the implementation of content-based image retrieval in large product image database. We present a two-stage product image retrieval method, with fully consideration of individual features of product images. In the initial pruning stage, shape feature based on salient edges of product object is used to generate a moderate number of candidates; in the second stage, the proposed detail feature combined with color and texture features is used for fully retrieval. Experiments show that this two-stage retrieval method accelerates search process with a high accuracy.

Prediction of Column Axial Forces in Inverted V-braced Seismic Steel Frames Considering Brace Buckling

Brace buckling in inverted-V-braced frames induces the vertical unbalanced force. The columns in the braced bay should be designed, per the capacity design concept, to remain elastic for gravity load actions and additional column axial forces that result from the brace buckling. However, owing to the difficulty in accumulating the buckling-induced column forces from different stories, empirical and often conservative approaches have been used in design practice. In this paper, three combination rules for a rational estimation of the column axial forces are proposed. The idea central to the three methods is to detect the stories with high buckling potential as precisely as possible by using pushover analysis and/or simple demand-to-capacity analysis. The vertical unbalanced forces in the stories detected as high buckling potential are then summed in a linear manner, whereas those otherwise are combined by following the SRSS (square root of sum of squares) rule. The accuracy and design advantage of the three methods is evaluated on the basis of extensive inelastic dynamic analyses. The mode shape-based method (MSBM), which is both simple and accurate, is recommended as the method of choice for practicing engineers among the three proposed.

Binding ligand prediction for proteins using partial matching of local surface patches.

Functional elucidation of uncharacterized protein structures is an important task in bioinformatics. We report our new approach for structure-based function prediction which captures local surface features of ligand binding pockets. Function of proteins, specifically, binding ligands of proteins, can be predicted by finding similar local surface regions of known proteins. To enable partial comparison of binding sites in proteins, a weighted bipartite matching algorithm is used to match pairs of surface patches. The surface patches are encoded with the 3D Zernike descriptors. Unlike the existing methods which compare global characteristics of the protein fold or the global pocket shape, the local surface patch method can find functional similarity between non-homologous proteins and binding pockets for flexible ligand molecules. The proposed method improves prediction results over global pocket shape-based method which was previously developed by our group.

Support vector machine–based facial-expression recognition method combining shape and appearance

Facial expression recognition can be widely used for various applications, such as emotion-based human-machine interaction, intelligent robot interfaces, face recognition robust to expression variation, etc. Previous studies have been classified as either shape- or appearance-based recognition. The shape-based method has the disadvantage that the individual variance of facial feature points exists irrespective of similar expressions, which can cause a reduction of the recognition accuracy. The appearance-based method has a limitation in that the textural information of the face is very sensitive to variations in illumination. To overcome these problems, a new facial-expression recognition method is proposed, which combines both shape and appearance information, based on the support vector machine (SVM). This research is novel in the following three ways as compared to previous works. First, the facial feature points are automatically detected by using an active appearance model. From these, the shape-based recognition is performed by using the ratios between the facial feature points based on the facial-action coding system. Second, the SVM, which is trained to recognize the same and different expression classes, is proposed to combine two matching scores obtained from the shape- and appearance-based recognitions. Finally, a single SVM is trained to discriminate four different expressions, such as neutral, a smile, anger, and a scream. By determining the expression of the input facial image whose SVM output is at a minimum, the accuracy of the expression recognition is much enhanced. The experimental results showed that the recognition accuracy of the proposed method was better than previous researches and other fusion methods.

Transfer Orbit Design and Control Based on Forbes' Assumption

In order to meet the requirements of space operation, the spacecrafts must maneuver agilely with the artifical propulsion. The orbit under artificial control is termed as Non-Keplerian orbit which does not follow Kepler’s Laws. The transfer orbit design under the continuous thrust is one of the most important topics in this new field. Shape-based method for Non-Keplerian orbit design is developed in this paper. Firstly, the equations of motion are established in polar coordinates system. And then the nondimensional variables are introduced for computation accuracy and speed, which give rise to nondimensional equations of motion. The general equation is derived with which common curves could be utilized in orbit design. In addition, the orbit design method is described based on Forbes’ velocity assumption and the formulation of the radius r with respect to time t, which is a sinusoidal function. The determination of the coefficients causes the orbit design problem to translate into an orbit control problem. The requisite thrust magnitude and direction are available via the simplified nondimensional equations of motion. In the end, an example of the transfer orbit is given. The result demonstrates that the shape-based method is feasible for the transfer orbit design or control problem under the continuous thrust, and the fuel expenditure is practicable.

Cervical spine image retrieval with semantic shape features

The objective of the proposed approach is to narrow down the semantic gap between the query and retrieval primitives in a spine image retrieval system. The proposed retrieval technique is based on a geometric eight-point model to enable formation of semantic query. The geometric eight-point model is uniform for all vertebrae, not biased by human experts and thus free from ambiguity. The features are extracted after automatically locating contour points of eight-point geometric model of spine vertebra. Then, vertebra is represented in feature space using region-based shape features, indicative of pathology (at two anterior corners). The proposed retrieval scheme uses Euclidean distance as similarity measure in feature space. It yields better result than an existing whole shape-based matching method for retrieval of spine images that uses Procrustes metric, in term of accuracy and parsimony of model. The approach is simple and computationally efficient.

3D level set reconstruction of model and experimental data in Diffuse Optical Tomography

The level set technique is an implicit shape-based image reconstruction method that allows the recovery of the location, size and shape of objects of distinct contrast with well-defined boundaries embedded in a medium of homogeneous or moderately varying background parameters. In the case of diffuse optical tomography, level sets can be employed to simultaneously recover inclusions that differ in their absorption or scattering parameters from the background medium. This paper applies the level set method to the three-dimensional reconstruction of objects from simulated model data and from experimental frequency-domain data of light transmission obtained from a cylindrical phantom with tissue-like parameters. The shape and contrast of two inclusions, differing in absorption and diffusion parameters from the background, respectively, are reconstructed simultaneously. We compare the performance of level set recons uction with results from an image-based method using a Gauss-Newton iterative approach, and show that the level set technique can improve the detection and localisation of small, high-contrast targets.

Shape-Based Active Contours for Fast Video Segmentation

In this letter, we propose a shape-based active contours method for segmentation, based on a piecewise-constant approximation of the Mumford-Shah (M-S) functional. The Chan-Vese (C-V) formalism in a level set framework is used to formulate our method; however no sign distance function (SDF) is employed in the method proposed here. This method has the topology-free segmentation associated with the C-V algorithm and adds faster convergence, less memory requirement and fast re-initialization. These properties make the algorithm very attractive for video segmentation.

Combining spatial support information and shape-based method for tomographic imaging inside a microwave cylindrical scanner

A nonlinear inverse scattering problem is solved to retrieve the permittivity maps inside a microwave cylindrical scanner of circular cross-section. In this article, we show how we can improve this minimization scheme by taking advantage of several a priori pieces of information. In particular, a global representation based on a Zernike basis expansion is introduced in order to restrain the class of solutions to functions which have circular spatial support, as is the case with the encountered geometrical configuration. The level-set function formalism is also exploited as the targets are known to be homogeneous by parts. We will show how we can combine the spatial support information and the binary nature of the scatterer, with limited changes of the inversion algorithm. Both synthetic and experimental results will be presented in order to highlight the importance of combining all the pieces of available information.

Hybrid health monitoring of prestressed concrete girder bridges by sequential vibration-impedance approaches

A hybrid health monitoring system using sequential vibration-impedance approaches is newly proposed to detect two damage types in prestressed concrete (PSC) girder bridges, which are tendon damage and girder damage. It mainly consists of three sequential steps: global damage warning, damage classification, and damage estimation. In Step 1, a global occurrence of damage is warned by monitoring changes in acceleration features. In Step 2, damage-types are classified into either tendon damage or girder damage by recognizing patterns of vibration and impedance features. In Step 3, the location and the extent of damage are estimated by using a mode shape-based method and a frequency-based method. The feasibility of the proposed system is evaluated on a laboratory-scaled PSC girder model for which acceleration responses and electro-mechanical impedances were measured for several damage scenarios of tendon damage and girder damage.

Best linear unbiased axial localization in three-dimensional fluorescent bead tracking with subnanometer resolution using off-focus images

A three-dimensional particle tracking technique, based on microscope off-focus images, was introduced in Z. Zhang and C.-H. Menq, Appl. Opt.47, 2361 (2008) and applied to bright-field imaging. This paper presents two major improvements to the axial localization algorithm of the 3D particle tracking technique. First, it extends the algorithm to measure fluorescent particles in the presence of photobleaching and excitation variation. Second, it enhances the measurement resolution by achieving the best linear unbiased estimation of the particle's axial position. Similarly to the original algorithm, a radius vector is first converted from the off-focus 2D image of the particle, and the axial position is estimated by comparing the radius vector with an object-specific model, calibrated automatically prior to each experiment. Although it was an intensity-based method, by normalizing the radius vectors the improved algorithm becomes a shape-based method, thus invariant to image intensity change and robust to photobleaching. Moreover, when considering the noise variance of each point in the radius vector and their correlations, the best linear unbiased estimation based on a linearized model is achieved. It is shown that variance equalization and correlation-weighted optimization greatly reduce the estimation variance and lead to near-uniform localization resolution over the entire measurement range. Estimation resolution is theoretically analyzed and validated by experiments. Theoretical analysis enables the prediction of measurement resolution based on calibration data. Finally, experimental results are presented to illustrate the performance of the measurement method in terms of measurement precision and range, as well as its robustness to intensity variation.

Molecular docking studies on DMDP derivatives as human DHFR inhibitors

Molecular docking is routinely used for understanding drug-receptor interaction in modern drug design. Here, we describe the docking of 2, 4-diamino-5-methyl-5-deazapteridine (DMDP) derivatives as inhibitors to human dihydrofolate reductase (DHFR). We docked 78 DMDP derivates collected from literature to DHFR and studied their specific interactions with DHFR. A new shape-based method, LigandFit, was used for docking DMDP derivatives into DHFR active sites. The result indicates that the molecular docking approach is reliable and produces a good correlation coefficient (r² = 0.499) for the 73 compounds between docking score and IC(50) values (Inhibitory Activity). The chloro substituted naphthyl ring of compound 63 makes significant hydrophobic contact with Leu 22, Phe 31 and Pro 61 of the DHFR active site leading to enhanced inhibition of the enzyme. The docked complexes provide better insights to design more potent DHFR inhibitors prior to their synthesis.

Ball route estimation under heavy occlusion in broadcast soccer video

This paper deals with analysis of broadcast soccer video. Estimation of ball movements is necessary to recognize interesting scenes such as goal scenes. It is, however, sometimes difficult to detect a ball using a simple color-based or shape-based method when the ball overlaps with multiple players and lines. We therefore developed a method of estimating a ball route during such overlaps by considering spatio-temporal relationships among players, lines, and the ball. The method can deal with difficult cases such as one in which a ball disappears at a player and reappears from another player. Experimental results demonstrate the effectiveness of the method.

SHAPE-BASED INDIVIDUAL/GROUP DETECTION FOR SPORT VIDEOS CATEGORIZATION

We present a shape-based method for automatic people detection and counting without any assumption concerning camera motion. In order to evaluate the robustness of the proposed method, we apply it for classifying athletics videos into two classes: videos of individual and videos of team sports. The videos used are real and characterized by dynamic and unconstrained environment. Moreover, in the case of team sport, we propose a shape deformations based method for running/hurdling discrimination (activity recognition). Robust, adaptive and independent from color, illumination changes and the camera motion, the proposed features are combined in the Transferable Belief Model (TBM) framework providing a two-level (frames and shot) video categorization. Experimental results of 97% of accuracy for individual/team sport categorization using a dataset of 252 real videos of athletic meetings, acquired by moving cameras under varying view angles, indicate the stability and the good performance of the proposed scheme.

Quantification and visualization of three-dimensional inconsistency of the ventrointermediate nucleus of the thalamus in the Schaltenbrand-Wahren brain atlas

This work quantifies and visualises 3D inconsistencies of the ventrointermediate nucleus (VIM) of the thalamus, including the VIM externum (VIMe) and VIM internum (VIMi), in the Schaltenbrand-Wahren (SW) brain atlas. For each VIM, VIMe, VIMi the 3D models, 3D-A, 3D-C and 3D-S were reconstructed from the SW axial, coronal and sagittal microseries, respectively, by applying a shape-based method. All 3D models, placed in the SW coordinate system, were compared quantitatively in terms of location (centroids), size (volumes), shape (normalised eigen values), orientation (eigen vectors), and mutual spatial relationships (overlaps and inclusions). The reconstructed 3D models differ significantly in location, size, shape, and inclusion rate. The centroid of 3D-A/VIM differs considerably from those of 3D-C/VIM and 3D-S/VIM. The difference between the centroids of 3D-C/VIM and 3D-S/VIM is in laterality only: that of 3D-C/VIM is located more medially (11.85 mm) than that of 3D-S/VIM (14.62 mm). 3D-A/VIM has the smallest volume (69.00 mm(3)); 3D-C/VIM is 3.71 and 3D-S/VIM 3.89 times larger. The overlap is also highly variable: 104.88 mm(3) for 3D-C/VIM with 3D-S/VIM, and very low (3.22 and 7.45 mm(3)) when 3D-A/VIM is involved. The highest inclusion rate is for 3D-C/VIM with 3D-S/VIM (39.10 and 40.97%) and the lowest for 3D-A/VIM with 3D-C/VIM (1.26 and 4.66%). The centroid of 3D-A/VIMe differs noticeably from those of 3D-C/VIMe and 3D-S/VIMe. The difference between the centroids of 3D-C/VIMe and 3D-S/VIMe is mainly in laterality: that of 3D-C/VIMe is located more medially (12.91 mm) than that of 3D-S/VIMe (16.65 mm). 3D-A/VIMe has the smallest volume (49.87 mm(3)); 3D-S/VIMe is 3.24 and 3D-C/VIMe 3.36 times larger. The overlap sizes are low: 32.72 mm(3) for 3D-C/VIMe with 3D-S/VIMe, and very low (1.32 and 2.01 mm(3)) when 3D-A/VIMe is involved. The inclusion rates are also low: the highest is for 3D-C/VIMe with 3D-S/VIMe (19.53 and 20.29%) and the lowest for 3D-A/VIMe with 3D-C/VIMe (1.19 and 4.01%). Lateral scaling of the coronal microseries by 1.2897 to match the 3D-C/VIMe and 3D-S/VIMe centroids increases the inclusion rates for the sagittal microseries by more than twice. The volume of scaled 3D-C enlarges to 216.24 mm(3) which is 1.34 bigger than that of 3D-S. There are substantial differences among the centroids of 3D-A/VIMi, 3D-C/VIMi and 3D-S/VIMi. The centroid of 3D-A/VIMi is located more anteriorly (-1.92 mm) than that of 3D-C/VIMi (-5.02 mm). The centroid of 3D-A/VIMi is located more ventrally (2.88 mm) than those of 3D-C/VIMi and 3D-S/VIMi (each at 5.34 mm). 3D-A/VIMi has the smallest volume (19.75 mm(3)); 3D-S/VIMi is 3.23 and 3D-C/VIMi 4.30 times larger. 3D-A/VIMi practically does not overlap with 3D-C/VIMi and 3D-S/VIMi. The inclusion rates for 3D-C/VIMi with 3D-S/VIMi are medium (32.63 and 43.43%). Each VIM, VIMe, VIMi as reconstructed from the SW atlas has a significant 3D inaccuracy within each orientation and across them. Therefore, absolute and direct reliance on the original SW atlas is unreliable and unsafe, and this atlas has to be used with great care and understanding of its strengths and limitations.

Shape-based ultrasound tomography using a Born model with application to high intensity focused ultrasound therapy

A shaped-based ultrasound tomography method is proposed to reconstruct ellipsoidal objects using a linearized scattering model. The method is motivated by the desire to detect the presence of lesions created by high intensity focused ultrasound (HIFU) in applications of cancer therapy. The computational size and limited view nature of the relevant three-dimensional inverse problem renders impractical the use of traditional pixel-based reconstruction methods. However, by employing a shape-based parametrization it is only necessary to estimate a small number of unknowns describing the geometry of the lesion, in this paper assumed to be ellipsoidal. The details of the shape-based nonlinear inversion method are provided. Results obtained from a commercial ultrasound scanner and a tissue phantom containing a HIFU-like lesion demonstrate the feasibility of the approach where a 20 mm x 5 mm x 6 mm ellipsoidal inclusion was detected with an accuracy of around 5%.

Novel Approach to Structure-Based Pharmacophore Search Using Computational Geometry and Shape Matching Techniques

Computationally efficient structure-based virtual screening methods have recently been reported that seek to find effective means to utilize experimental structure information without employing detailed molecular docking calculations. These tools can be coupled with efficient experimental screening technologies to improve the probability of identifying hits and leads for drug discovery research. Commercial software ROCS (rapid overlay of chemical structures) from Open Eye Scientific is such an example, which is a shape-based virtual screening method using the 3D structure of a ligand, typically from a bound X-ray costructure, as the query. We report here the development of a new structure-based pharmacophore search method (called Shape4) for virtual screening. This method adopts a variant of the ROCS shape technology and expands its use to work with an empty crystal structure. It employs a rigorous computational geometry method and a deterministic geometric casting algorithm to derive the negative image (i.e., pseudoligand) of a target binding site. Once the negative image (or pseudoligand) is generated, an efficient shape comparison algorithm in the commercial OE SHAPE Toolkit is adopted to compare and match small organic molecules with the shape of the pseudoligand. We report the detailed computational protocol and its computational validation using known biologically active compounds extracted from the WOMBAT database. Models derived for five selected targets were used to perform the virtual screening experiments to obtain the enrichment data for various virtual screening methods. It was found that our approach afforded similar or better enrichment ratios than other related methods, often with better diversity among the top ranking computational hits.

A Shape-Based Buffering Method

Distance constraint is a major concern in many spatial analyses. Buffering is one of the proximity techniques in GIS most commonly used to address this constraint. I introduce shape-based point buffering, an anisotropic and variable-distance buffer generation method conformal to the original polygons. In contrast with isotropic fixed-distance buffering, shape-based buffering is defined using a relative distance (percentage) instead of a real unit (for example, meters), and it allows all buffered boundaries to be formed at the same time. The construction and implementation of the buffering method are described below. Three emergency-response scenarios are designed to demonstrate potential applications of this buffering method, including a shape-based fixed-percentage buffer calculation and space distribution, a shape-based variable-percentage buffer region, and a reverse calculation of the task schedule from a probability surface constructed from shape-based buffers. Limitations of the method are discussed. The method has potential applications in emergency preparedness and planning to better address fairness issues when a geographic area must be zoned arbitrarily.

An analysis of facial expression recognition under partial facial image occlusion

In this paper, an analysis of the effect of partial occlusion on facial expression recognition is investigated. The classification from partially occluded images in one of the six basic facial expressions is performed using a method based on Gabor wavelets texture information extraction, a supervised image decomposition method based on Discriminant Non-negative Matrix Factorization and a shape-based method that exploits the geometrical displacement of certain facial features. We demonstrate how partial occlusion affects the above mentioned methods in the classification of the six basic facial expressions, and indicate the way partial occlusion affects human observers when recognizing facial expressions. An attempt to specify which part of the face (left, right, lower or upper region) contains more discriminant information for each facial expression, is also made and conclusions regarding the pairs of facial expressions misclassifications that each type of occlusion introduces, are drawn.