Search

Cell segmentation is challenging owing to the existence of various experimental configurations, cell shapes that cannot be mathematically defined, and ambiguous cell boundaries. We propose a cell segmentation method using cell region discriminator and multi-cell discriminator trained using heterogeneous machine-learning techniques such as logistic regression, expectation-maximization, and support vector machine (SVM). The cell-region discriminator identifies the regions where cells are found in images obtained from microscopes via a secondary logistic regressor, and its features use statistical information as well as the distribution of neighbor intensities. The SVM-based multi-cell discriminator determines whether multiple cells are present in the region detected by the cell-region discriminator and whether the region should be divided using the expectation-maximization algorithm. We suggest features for the boundary sectional area and the least square error for cell surface fitting to train the multi-cell discriminator. Using the features and the SVM, the multi-cell discriminator can be trained without overfitting, even for small training data. During this process, the proposed convex cell surface enhances the clustering performance. In experiments, our method based on two discriminators stably divided connected cells even when the contrast between a cell and the background area was small, and it outperformed state-of-the-art methods in terms of cell detection and segmentation accuracy.

The inaccuracy in Hounsfield unit (HU) of dental cone beam computed tomography (CBCT) systems has been reported in several previous studies. CT images produced by dental CBCTs cannot represent an object's density owing to the HU inaccuracy; therefore, dental surgeons do not trust or utilize the HU values from CBCT in dental surgical operations, especially in implant placement. To provide more accurate bone density, we considered the possibility of a dual energy (DE) imaging technique. A new dental CBCT system capable of switching kVps was developed. The projection images at different kVp were reconstructed independently; subsequently, a new set of CT images of more accurate HUs would be produced. Moreover, a new calibration procedure was proposed in addition to the conventional HU calibration. Cylindrical hydroxyapatite (HA) phantoms of different densities were developed for this calibration. The voxel values of CT images at different energies were correlated point-by-point, and the parameters for mapping more accurate HUs values were extracted in this calibration. The proposed calibration and 30 examinations were performed using an acrylic base with cylindrical HA phantoms of different densities. The differences between the HUs of resultant DE images and the CT values provided by the phantom manufacturer were approximately 1%, and the coefficients of variation (CV) from 30 repeated examinations were near 1%. The HUs from our new DE CBCT agreed well with the CT values in this phantom study.

We propose a technique to enhance the accuracy of the Bayesian classifier for frontal lobe T1 — weighted images. In order to overcome the difference in intensity between different objects, we conducted experiments on the discrimination of likelihood and the relation between the upper error limit and the classification accuracy of contrast based stretching. The Naive Bayesian classifier was used for verification in consideration of the low prior probability and the final experiment using the cross validation method of the approximate model, showing three classes (gray matter, white matter, cerebrospinal fluid). The purpose of this study is to provide a preliminary study to assist in the analysis of lesions and patterns of frontal lobe MR imaging.

Accurate cell segmentation is one of the critical, yet challenging problems in microscopy images due to ambiguous boundaries as well as a wide variation of shapes and sizes of cells. Even though a number of existing methods have achieved decent results for cell segmentation, boundary vagueness between adjoining cells tended to cause generation of perceptually inaccurate segmentation of stained nuclei. We propose a segmentation method of cells based on structural dissimilarity between embodied and imaged cells. From assumption that the shape of the region of adjoining cells follows a 2D Gaussian mixture model, the cell region is divided by an expectation-maximization method. The lowest structural dissimilarity using embodied cells decides on the number of components of the 2D Gaussian mixture model. The region of interest is extracted by implementation of both global and local thresholdings, which performs binarization of the local image with a seed at the center, where the seed is obtained by the maximally stable extremal regions. Our approach presented considerably higher evaluation scores compared with other five existing methods in terms of both accuracies of region of interest (ROI) detection and boundary discrimination.

Existing maximum-margin support vector machines (SVMs) generate a hyperplane which produces the clearest separation between positive and negative feature vectors. These SVMs are effective when datasets are large. However, when few training samples are available, the hyperplane is easily influenced by outliers that are geometrically located in the opposite class. We propose a modified SVM which weights feature vectors to reflect the local density of support vectors and quantifies classification uncertainty in terms of the local classification capability of each training sample. We derive a primal formulation of an SVM that incorporates those modifications, and implement an RC-margin SVM of the simplest form. We evaluated our model on the recognition of handwritten numerals, and obtained higher recognition rate than a standard maximum-margin SVM, a weighted SVM, or an SVM which accounts for classification uncertainty.

User-based collaborative filtering suffers from significant amount of computational overhead to find users similar to a target user. To reduce the overhead, we propose a novel method to identify unnecessary users and items in computing the similarity. Also, we propose a data structure to support the method quite efficiently. Through extensive experiments, we show the proposed method outperforms traditional methods up to 33.8 times.

Quantitative analysis of nanostructures from scanning electron microscope (SEM) images requires a clear segmentation of grains and their boundaries. This is not provided by active contour models, which also require user guidance. Our automatic technique creates a rough representation of grain boundaries by adaptive thresholding. It then performs raycasting from a rectangular grid of seed points to ensure that the grain shapes are convex, and selects the best result for each grain. The whole process can be repeated several times to improve the segmentation. We present results for images of titanium foil, which show that our approach compares favorably in terms of speed and segmentation quality with four competing techniques.

We propose smart backpack to guide to destination safely for visually impaired person. To estimate the current position of the blind is very important. We use Zigbee modules that estimate the distance using the signal strength in transmission and reception period. We build a grid-based local area map for guiding user. In addition, we use ultrasonic sensors for detecting obstacles and analysing the spatial environment in front of user. We make a backpack type wearable system by integrating these technologies for the convenient outdoor walking.

The Android platform used for mobile devices can be applied to embedded systems, such as robot control systems. Developers should create applications by using Java language provided by Android SDK for embedded systems operated via Android platforms. However, in many existing embedded systems, developers have written applications for controlling the system by using C language. Android NDK makes it possible for developers to easily reuse such legacy code written in C/C++ languages. In this paper, we show the difference in terms of performance between an Android application using native code library from C source and an Android application using the same algorithm written in Java language only. We conducted an experiment on five parts: JNI communication delay, integer calculation, floating-point calculation, memory access algorithm, and heap memory allocation algorithm. This paper presents a guideline for an effective way to use native code libraries in Android applications.

A group of documents is called near-duplicates if they are almost the same with just a slight difference. Since near-duplicates are major concerns of Web search engines, it is necessary to identify and filter them effectively. Among existing near-duplicate identification methods, MinHashing is the most well-known one. It identifies near-duplicates regardless of locations of different parts in two documents. In blog environment, however, most near-duplicates differ only in their beginning or end. According to our preliminary experiment, about 99% of near-duplicates differ in the beginning or end (blog-duplicates hereafter) and only 1% of them differ in the middle. Thus, blog-duplicates have a long matched sequence in their central parts. Based on this important observation, we present a novel algorithm, Central Match, to identify blog-duplicates efficiently and accurately. When searching a document database for possible log-duplicates of a given document, Central Match runs50 times faster than MinHashing. In addition, Central Match identifies blog-duplicates more accurately than MinHashing. According to our experiments, when the precisions of Min-Hashing and Central Match are fixed to 0.9, their recalls are around 0.5 and 0.9, respectively, which means Central Match finds 80% more blog-duplicates than MinHashing.

The road network now opens a new application area for the classic k-NN queries, which retrieve k objects closest to a given query point. There are many previous methods for k-NN query processing. However, all of them are not sufficient in experimental evaluation of performance. In this paper, we present the way that extensively evaluates prior methods for processing of k-NN queries. To achieve this goal, we first describe well-known previous methods for retrieving k objects closest to a given query point. Next, we establish the criteria against which previous methods have been evaluated for k-NN query processing. Last, we build the experiment environment using the real network data and then perform extensive performance evaluation.

We have designed a framework for content based appraisal of documents. Our motivation is to provide computer assisted support for answering several appraisal criteria according to the general appraisal guidelines in the National Archives and Record Administration (NARA) 1441 directive. The appraisal criteria led us to investigations related to (a) finding groups of PDF documents with similar content, (b) ranking documents according to their creation/ modification time and digital volume, and (c) detecting inconsistency between ranking and content within a group of related documents. The novelty of our work is in designing a methodology and a mathematical framework for document appraisals, and prototyping the framework working with text, image and vector graphics components of PDF documents. We present example results of grouping, ranking and integrity verification for groups of scientific documents about medical topics.

This paper addresses the problem of understanding preservation and reconstruction requirements for computer- aided medical decision-making. With an increasing number of computer-aided decisions having a large impact on our society, the motivation for our work is not only to document these decision processes semi-automatically but also to understand the preservation cost and related computational requirements. Our objective is to support computer-assisted creation of medical records, to guarantee authenticity of records, as well as to allow managers of electronic medical records (EMR), archivists and other users to explore and evaluate computational costs (e.g., storage and processing time) depending on several key characteristics of appraised records. Our approach to this problem is based on designing an exploratory simulation framework for investigating preservation tradeoffs and assisting in appraisals of electronic records. We have a prototype simulation framework called image provenance to learn (IP2Learn) to support computer-aided medical decisions based on visual image inspection. The current software enables to explore some of the tradeoffs related to (1) information granularity (category and level of detail), (2) representation of provenance information, (3) compression, (4) encryption, (5) watermarking and steganography, (6) information gathering mechanism, and (7) final medical report content (level of detail) and its format. We illustrate the novelty of IP2Learn by performing example studies and the results of tradeoff analyses for a specific image inspection task.

Fiber-reinforced composites are used as insulating layers in PCBs to reduce thermal residual stress and warpage due to their low coefficient of thermal expansion (CTE) and high elastic modulus. The composite prepreg is laminated on patterned copper layers by hot-press molding process to apply temperature and pressure cycles. In this study, the lamination process of glass/epoxy woven composite in the chip- embedded PCB was investigated in order to manufacture reliable chip-embedded PCB. Temperature profiles in several panels stacked in the hot-press were measured to estimate degree of cure and resin-fill characteristics. The degree of cure was evaluated by a cure kinetic model constructed from dynamic and isothermal differential scanning calorimetry (DSC), while the resin flow through via holes and around embedded chips was analyzed by a viscosity model constructed from isothermal rheometry.

Cord blood mononuclear cells (CBMC) are responsible for neonatal host defense against microbial infections. Among them, monocytes and neutrophils are first-line of host defense cells in bacterial infections. Previously, we reported that herpes virus entry mediator protein (HVEM) ligation by its ligand, lymphotoxin-like inducible protein that competes with glycoprotein D for binding herpes virus entry mediator on T cells (LIGHT; TNFSF14), stimulates adult peripheral blood (PB) monocytes and neutrophils to increase their bactericidal activities. In this study, we investigated the role of HVEM on cord blood (CB) monocytes and neutrophils. HVEM expression was significantly lower on CB B- and T-lymphocytes than PB B- and T-lymphocytes. However, HVEM expression on CB monocytes and neutrophils was similar to PB monocytes and neutrophils. Moreover, LIGHT treatment to CB monocytes and neutrophils enhanced CC chemokine receptor 1 (CCR1) and intracellular adhesion molecule-1 (ICAM-1, CD54) expressions, which are used for cell migration during bacterial infections. These results suggest CB HVEM activation is important in neonatal host defense against bacterial infections.

Most of micro-CTs are capable of micron-resolution imaging if a small sample is of interest. Recent micro-CT developments with a flat-panel detector make it possible to scan the whole body of a small animal as large as a medium sized rat. However, the spatial resolution in the whole body imaging should be sacrificed down to the order of 100 mum due to the limitation in the number of X-ray detector pixels and low signal-to-noise ratio. In this study, we introduce an X-ray zoom-in micro tomography system which has localized imaging capability, i.e., high resolution imaging of a local ROI inside a large subject, for small animal imaging studies. In zoom-in micro tomography, two projection data sets, one obtained with a full FOV scan and the other obtained with a limited FOV scan, are acquired with different magnification ratios. To obtain exact images of the ROI, the out-of-FOV projection data are calculated from the full FOV projection data, and then, the calculated projection data are augmented to the limited FOV projection data. The zoom-in micro tomography system consists of a micro focus X-ray source, a CMOS flat-panel detector, a rotating gantry and a parallel data processing system. We have used the zoom-in micro tomography system for in vivo femoral bone imaging in osteoporosis studies with SD rat models in which fine trabecular bone structure imaging is essential.

In this paper, we address the problem of 3D volume reconstruction from depth adjacent subvolumes (i.e., sets of image frames) acquired using a confocal laser scanning microscope (CLSM). Our goal is to align sub-volumes by estimating an optimal global image transformation which preserves morphological smoothness of medical structures (called features, e.g., blood vessels) inside of a reconstructed 3D volume. We approached the problem by learning morphological characteristics of structures inside of each sub-volume, i.e. centroid trajectories of features. Next, adjacent sub-volumes are aligned by fusing the morphological characteristics of structures using extrapolation or model fitting. Finally, a global sub-volume to subvolume transformation is computed based on the entire set of fused structures. The trajectory-based 3D volume reconstruction method described here is evaluated with a pair of consecutive physical sections using two evaluation metrics for morphological continu

A dual-modality compact SPECT/CT system for small animal imaging has been developed. The SPECT system consisted of a pinhole collimator and plate type NaI(Tl) scintillation crystal coupled to a PSPMT. The X-ray system consisted of a microfocus X-ray tube and a CMOS flat-panel detector. The SPECT system was mounted perpendicular to the X-ray system. Individual projections of SPECT and CT were acquired by rotating the animal about a vertical axis in front of the detectors. The SPECT and CT image were reconstructed using OSEM and Feldkamp's cone-beam algorithms, respectively. Mouse and rat SPECT demonstrated detailed activity distributions at the expected structures. A CT image obtained with 40 RVp and 0.5 mA presented high-resolution anatomic details. Fused SPECT/CT images demonstrated good agreement between the CT images and the corresponding uptake of the radiotracer. The SPECT/CT system developed in this study provides high-quality dual-modality images and could be useful to obtain functional images with high-resolution morphology information.

To compete successfully in today's global online game markets, a cross national analysis is becoming a more important issue. The purpose of this research is to identify if causalities among flow and customer loyalty are different between Korean and Japanese online games. To perform our research, we use Multi- group Structural Equation Model.

In this paper, we introduce an X-ray microtomography system capable of high-resolution zoom-in imaging of a local region inside a large object. By combining two kinds of projection data, one from the full field-of-view (FOV) scan of the whole object and the other from the limited FOV scan of the region of interest, we have obtained zoomed-in images of the region of interest. With computer simulations, we have found that the proposed zoom-in imaging technique has superb spatial resolution, comparable to that of local tomography, without any contrast anomalies commonly appearing in conventional local tomography. To verify the zoom-in imaging capability experimentally, we have integrated a micro tomography system using a micro-focus X-ray source, a flat-panel CMOS X-ray detector with the matrix size of 1248times1248, and a precision scan mechanism. The mismatches between the two projection data caused by misalignments of the scan mechanism have been estimated with a calibration phantom and the mismatch effects have been compensated in the image reconstruction procedure. Zoom-in imaging results of bony tissues with the spatial resolution of 20 mum suggest that zoom-in microtomography can be greatly used for high resolution imaging of a local region in small animal imaging studies

We propose a new methodology for fusing temporally changing multisensor raster and vector data by developing a spatially and temporally varying uncertainty model of acquired and transformed multisensor measurements. The proposed uncertainty model includes errors due to (1) each sensor by itself, e.g., sensor noise; (2) transformations of measured values to obtain comparable physical entities for data fusion and/or to calibrate sensor measurements; (3) vector data spatial interpolation that is needed to match different spatial resolutions of multisensor data; and (4) temporal interpolation that has to take place if multisensor acquisitions are not accurately synchronized. The proposed methodology was tested using simulated data with varying (a) amount of sensor noise, (b) spatial offset of point sensors generating vector data, and (c) model complexity of the underlying physical phenomenon. We demonstrated the multisensor fusion approach with a data set from a structural health monitoring application domain.

Since the early 2000's, the online games have been recognized as a major profit model within e-commerce and have been developed into the core of the world cultural industries. However, Chinese online game industry has encountered higher competition in global market. To survive successfully in today's competitive online game markets, Chinese online game companies need to determine what motivates their loyal customers. However, previous research didn't focus on identification of critical variables for market segmentation in online game market. The purposes of our research are to identify causalities among flow and customer loyalty in online games, and to identify the factors by which flow are influenced.

To survive successfully in today's competitive online game markets, Chinese online game companies need to determine who the target customers are and what motivates them. However, previous research didn't focus on identification of critical variables for market segmentation in online game market. Additionally, the traditional clustering methodologies have not provided a unique clustering nor determined the precise number of clusters. The purpose of our research is to identify the critical variables and to develop a new methodology for market segmentation of online game market. Our research tested the model with 438 Chinese online game users

Both the engagement of the T cell receptor with MHC-Ag complex and a second costimulatory signal are required for optimal T cell activation. Since T cells become nonreactive or tolerant without the second signal, blocking the appropriate costimulatory signal provided by APCs and T cells has been successful transplantation of allogeneic organs. We investigated the role of 4-1BB, an important costimulatory molecule for T cells, in alloimmune responses. When the 4-1BB costimulatory pathway was blocked, it led to a pronounced prolongation of allogeneic vascularized cardiac grafts. Thus, our data provide evidence that 4-1BB is required for alloimmune responses and present the possibility that blockade of the 4-1BB costimulatory pathway may be used as a therapeutic tool in preventing cardiac allograft rejection.

The authors propose a modified flooding algorithm with minimum first-derivative length (MFDL) path for a tactical circuit-switched grid network with a maximum of four links and two priority classes of voice traffic. They consider the minimization of average end-to-end blocking probability, and study the blocking performance of the modified flooding algorithm and the flood search algorithm. They also suggest a decision rule for the search message propagation scheme and analyze the signaling traffic load of the algorithm. It is found that with the same or lower cost of switch processor loading and call set-up time as compared to the flood search algorithm, the modified flooding algorithm provides the shortest path on the basis of blocking probability. Therefore, its use can improve the blocking performance of a circuit-switched network. >

We have applied a flat-panel detector to an X-ray cone-beam micro computed tomography (micro-CT) for small-animal imaging. The flat-panel detector consists of an active matrix of transistors and photodiodes with a pixel pitch of 50 /spl mu/m and a thallium-doped cesium iodide (CsI:Tl) scintillator as an X-ray-to-light conversion layer. The detector was fabricated with a CMOS (complementary metal-oxide-semiconductor) technology capable of a sub-micrometer design line width, hence, it has a pixel fill-factor as high as /spl sim/80%. In addition, the detector has a very fast response characteristic with an image lag being less than 0.3% in the frame integration time of 5 s. Characterization of the CMOS flat-panel detector has been performed in terms of modulation transfer function (MTF), noise power spectrum (NPS), and detective quantum efficiency (DQE). Tomographic imaging performances of the micro-CT system, such as, voxel noise, contrast-to-noise ratio (CNR), and spatial resolution, have also been evaluated by using various quantitative phantoms. Experimental results of euthanized laboratory rat imaging suggest that the micro-CT system employing a CMOS flat-panel detector can be greatly used in small-animal imaging.

A hybrid direct-sequence slow-frequency-hopped (DS/SFH) spread-spectrum system is used to achieve a larger gain in some applications. The probability of error for a hybrid DS/SFH spread-spectrum system is calculated over a channel which suffers from multiple-tone jamming and additive white Gaussian noise (AWGN). Binary phase-shift keying (BPSK) is used for data modulation. The probability of error is computed for various system parameters, such as DS/SFH hybrid ratio, the number of jamming tones, a jamming-to-signal ratio, and a signal-to-noise ratio. It is shown that there are hybrid ratios with which a hybrid DS/SFH system achieves smaller probability of error than a DS system and a SFH system achieve for some values of system parameters. >