Search

Background Our aim was to demonstrate that automated detection and classification of breast microcalcifications, according to Breast Imaging Reporting and Data System (BI-RADS) categorisation, can be improved with the subtraction of sequential mammograms as opposed to using the most recent image only. Methods One hundred pairs of mammograms were retrospectively collected from two temporally sequential rounds. Fifty percent of the images included no (BI-RADS 1) or benign (BI-RADS 2) microcalcifications. The remaining exhibited suspicious findings (BI-RADS 4-5) in the recent image. Mammograms cannot be directly subtracted, due to tissue changes over time and breast deformation during mammography. To overcome this challenge, optimised preprocessing, image registration, and postprocessing procedures were developed. Machine learning techniques were employed to eliminate false positives (normal tissue misclassified as microcalcifications) and to classify the true microcalcifications as BI-RADS benign or suspicious. Ninety-six features were extracted and nine classifiers were evaluated with and without temporal subtraction. The performance was assessed by measuring sensitivity, specificity, accuracy, and area under the curve (AUC) at receiver operator characteristics analysis. Results Using temporal subtraction, the contrast ratio improved ~ 57 times compared to the most recent mammograms, enhancing the detection of the radiologic changes. Classifying as BI-RADS benign versus suspicious microcalcifications, resulted in 90.3% accuracy and 0.87 AUC, compared to 82.7% and 0.81 using just the most recent mammogram (p = 0.003). Conclusion Compared to using the most recent mammogram alone, temporal subtraction is more effective in the microcalcifications detection and classification and may play a role in automated diagnosis systems.

Breast cancer is one of the deadliest malignancies worldwide. In mammography, the most reliable screening tool for its diagnosis, expert radiologists review the mammograms to determine whether the patient has any signs of disease. Unfortunately, the evaluation of breast abnormalities is challenging, even for experienced radiologists. Computer-Aided Detection (CAD) systems can assist in the detection of breast cancer. In this work, an algorithm for the automatic detection and classification of masses, based on subtraction of sequential digital mammograms, image registration and machine learning, is presented. Previous studies assessed the use of sequential mammograms to perform temporal analysis by creating a new temporal feature vector. Temporal subtraction registers and subtracts the prior mammogram from the current one, prior to performing mass detection and classification. A new dataset, which includes sequential pairs from 40 patients (160 mammograms) with precisely annotated mass locations (benign and suspicious), was created to assess the performance of the algorithm. For the classification, various features were extracted and six classifiers were used in a leave-one-patient-out cross-validation. The accuracy of the classification of masses as benign or suspicious increased from 90.83% (with the previously described temporal analysis) to 96.51% (with temporal subtraction). The improvement was statistically significant with p < 0.05. These results demonstrate the effectiveness of the proposed technique of temporal subtraction of mammograms for the detection of masses.

We consider a new workload allocation policy addressing fairness for user level performance measures. More specifically the criterion used for optimal workload allocation is the one which minimizes the maximum expected response time at computer systems to which jobs are routed. The policy to attain this criterion is therefore referred to as the min-max policy (MMP). It is shown that this optimization criterion is tantamount to routing to the fastest M processors, where M depends on system statistics and equalizing the expected response times on these processors. The algorithm to compute job routing probabilities is applicable to increasing continuous functions of system response time versus the job arrival rate. We next investigate some properties of the MMP and show that it results in minimizing the coefficient of variation of response time when the job processing times are exponentially distributed. We compare the MMP with the one that minimizes the mean overall response time. It is shown that the new policy attains fairness by equalizing the mean response times at different systems, at a tolerable increase in overall response time. Finally, we report on a sensitivity analysis with respect to changes in job arrival rate and errors in estimating this rate.

The aim of this paper is to briefly present the overall objectives and the expected outcome of an on-going effort concerning the design the implementation and the analysis of next generation intelligent energy systems based on Anticipatory control and a set of ICT emerging technologies and innovations.

Large-scale systems comprising of multiple heterogeneous entities are directly influenced by the interactions of their participating entities. Such entities, both physical and virtual, attempt to satisfy their objectives by dynamically collaborating with each other, and thus forming collective adaptive systems. These systems are subject to the dynamicity of the entities' objectives, and to changes to the environment. In this work we focus on the latter, i.e.\ on providing the means for entities in such systems to model, monitor and evaluate their perceived utility by participating in the system. This allows for them to make informed decisions about their interactions with other entities in the system. For this purpose we propose a utility-based approach for decision making, as well as an architecture that allows for the support of this approach.

Workloads in distributed database applications consist of queries and transactions. In order to address performance requirements, distributed transaction processing systems have to deal with two related issues: transaction routing and scheduling. Due to the distribution of data objects among nodes and the access cost incurred by remote accesses, efficient transaction routing is an important consideration for overall system performance. Another important consideration is workflow scheduling and routing. Workflows are complex units of work consisting of multiple, possibly interdependent, transactions. In this survey, we discuss a number of different transaction routing mechanisms and their performance.

Interdependency is a constant of business. Businesses form complex networks for the exchanging of goods and services. Service networks models represent the interconnections among companies, their parts and individuals in terms of services that are provided and consumed. Due to the size of nowadays service networks, their models tend to grow very large. The effective management of service network models requires methods for limiting the amount of data presented on the basis of what is needed through views. We identify a hierarchy of views on service network models, namely offering-centric views, participant views and multilateral views. The offering-centric views focus on one service that is offered by one participant. Participant views present the entirety of the data related to one participant. Multilateral views represent the data about a set of participants. Additionally, we investigate the correlations between these views and the mechanisms for aggregating and projecting them from each other and from service network models.

One of the main goals in service science is to find efficient ways to analyze and increment the value in a service network. The approach we propose in this paper is to increase the agility of the system in such a way that the network and the underlying business processes are able to spontaneously react to changes in the requirements or in the environment. This is done by making extensive use of knowledge transfer and intangible interactions among network participants. The final outcome is a transformation and analysis methodology that may be applied to a wide variety of service networks with the aim of finding possible reconfigurations for increasing customer satisfaction, reducing transaction risks, and therefore increasing the overall value of the network.

The authors propose and evaluate an efficient hierarchical clustering and allocation algorithm that drastically reduces the interprocess communications cost while observing lower and upper bounds of utilization for the individual processors. They compare the algorithm with branch-and-bound-type algorithms that can produce allocations with minimal communication cost, and show a very encouraging time complexity/suboptimality tradeoff in favor of the algorithm, at least for a class of process clusters and their random combinations which it is believed occur naturally in distributed applications. The heuristic allocation is well suited for a changing environment, where processors may fail or be added to the system and where the workload patterns may change unpredictably and/or periodically. >

Consumers increasingly use online services in their everyday life: when they drive a car, at home and at work. We address the problem of selecting for the consumer, the most valuable online service, according to the consumer value preferences. We combine our work for value analysis of service networks with our work on selecting services based on quality of service requirements. We show with examples (car manufacturer and its service system), how service delivery monitoring and service brokers can be used to improve consumer satisfaction, and discuss how this work can be extended in the future, to achieve viable and adaptive, dynamically evolving service ecosystems.

The Internet has catered for the transformation of traditional "stovepiped" service companies into global service networks fostering co-production of value to more effectively and efficiently satisfy the ever-growing demands of mundane customers. The catalyst of this change is the happenstance of Service Oriented Computing, which provides a natural distributed computing technology paradigm for implementing and evolving such highly distributed networks of autonomous trading partners with coordinate and cooperative actions. However, how to faithfully (re-)map service networks to business processes and service realizations and vice-versa is still partly terra incognita. In this paper, we introduce a semi-automatic model transformation approach for creating the abstract business processes that take place between trading partners from models of service networks, assuming limited human-involvement focused on selecting reusable transformation patterns. This approach is explored and validated using a realistic case study reflecting best practices in the telecommunications industry.

In today’s Information Systems, users present credentials with local significance, to be authenticated and gain access to internal functionality. Users have different login-password combinations for each online service, or even different credentials for different roles within a service. As a result they tend to make poor password choices that are easy to remember, or even repeat the same login-password information on different services. This poses security threats to service providers and a privacy risk for end-users. The solution is to shift to identity management systems. Such a system will issue a digital identity for every user and will be able to control the full life-cycle of these identities, from creation to termination. Another aspect of such a system is the single sign-on mechanism, whereby a single action of user authentication and authorization can permit the user to access multiple services. The benefits are improved security, accountability and privacy protection.

Now that the first wave of excitement on e-commerce has subsided, and after the sobering experience of the dot-com bubble burst, there is a growing understanding that e-commerce is not about a different way of doing commerce, and e-business is not about a different way of doing business. E-Business is about doing business – in a better, more competitive and productive way. To improve business, one has to transform business and the processes that it uses.

A novel approach to allocating and sharing communication and computational resources in a distributed system is described. The approach, which is based on concepts drawn from microeconomics, uses algorithms that are competitive rather than cooperative. The effectiveness of these concepts is demonstrated by describing an economy that improves the performance of a distributed system by implementing load balancing. In this economy, competition sets prices for the resources in the system. Jobs complete for the resources by issuing bids, and the resource allocation decisions are made through auctions held by the processors. The benefits of the method include limited complexity and algorithms that are intrinsically decentralized and modular. Simulation studies show that these economies achieve substantial performance benefits. >

Workload management algorithms for satisfying administration defined response time goals in transaction processing systems are presented. Each arriving transaction belongs to a predefined transaction class, and the system administrator defines an average response time goal for each transaction class. A dynamic transaction priority algorithm and a set of transaction routing algorithms for multiple processor transaction systems are given. These algorithms set priorities and route transactions based on the data objects accessed by the transactions, the current load of each system, and the goal satisfaction of the transaction classes. The algorithms have low overhead, and a detailed simulation study shows that they substantially improve goal satisfaction compared to previous work. >

A new approach to performing resource allocation in autonomous distributed computer systems is explored. As opposed to previous work which is based on interprocessor cooperation, the distributed system is modeled as a competitive society of microeconomic agents. The model is applied to the problem of managing distributed, replicated data objects. In this economy, jobs and transactions submitting read/write operations on data objects pay the processors to perform the operations. The processors use the revenue they earn to lease copies of data objects on which they can make a profit. The data objects themselves are active agents and set the prices for leasing copies. The economy is evaluated by a simulation study. The experiments show that the economy can substantially improve performance by varying the placement and number of copies of each data object. >

The paper presents the HERMES specification language, which is introduced to express scripts for complex activities involving coordination and collaboration. The language is part of a framework for network-centric applications, providing a two-level framework for expressing metadata. Service flow rules define communication channels between software components and actions triggered by run time events. The semantics of HERMES supports a model for dynamic and adaptive work sessions in open environments.

Workflow management systems are developed to support and automate the execution of business processes, by assigning tasks to agents according to the workflow specification. Current state-of-the-art workflow systems are mainly concerned with the routing and assignment of tasks, providing little support for tasks related to administration and management. This deficiency is exacerbated in dynamic open environments, where services provided and managed by multiple autonomous authorities need to be integrated. Such environments significantly stretch the assumptions underlying current state-of-the-art workflow system designs, especially as network-centric applications become widespread. We propose a framework, service level management, for addressing all system administration in an integrated manner in the context of a workflow execution model that allows for dynamic adaptation (reconfiguration) of work sessions in a dynamic and heterogeneous run-time environment. Service level management requires a comprehensive monitoring infrastructure. We present our design for such an infrastructure, which is part of our current work towards developing the Aurora architecture for supporting work sessions.

The paper presents a technique for performing dynamic goal-oriented buffer pool management for database management systems. To dynamically adjust the buffer pool sizes for the multiple buffer pools provided by database management systems is a complex constrained optimization problem. In the goal-oriented approach, the user specifies each buffer pool's random access response time goal and the total available number of buffers for all buffer pools. The problem is to dynamically expand or contract the buffer pool sizes based on the database workload to achieve these pre-defined response time goals for each buffer pool while maintaining the same total number of buffers in the database system. Our goal satisfaction algorithm monitors goal satisfaction of each buffer pool and periodically changes buffer pool sizes to improve goal satisfaction. The expansion and contraction process does not allocate new or free up existing virtual storage. We demonstrate that dynamic tuning can greatly improve buffer pool goal satisfaction through trace driven simulations.

User interface adaptations can be used to address several user interaction challenges in the development of digital library systems. This paper: (a) examines some of the intrinsic characteristics of digital library systems; (b) identifies some of the key Human–Computer Interaction (HCI) challenges in relation to design and architectural abstraction for user interfaces to digital library systems; and (c) develops an argumentation for adaptations in digital library systems. By drawing parallels to recent HCI research into adaptable and adaptive user interaction, the paper illustrates potential areas in which user interface adaptation can provide a useful technique for advancing the quality of human interaction with a digital library system.

The NSF Invitational Workshop on Distributed Information, Computation, and Process Management for Scientific and Engineering Environments (DICPM) brought together domain specialists from engineering and the ocean, atmospheric, and space sciences involved in the development and use of simulations of complex systems, and computer scientists working on distributed repositories, visualization, and resource management. The objective was to formulate directions for research efforts to facilitate effective collaboration and to help increase access to information and sharing of results and tools useful in large-scale, distributed, multidisciplinary scientific and engineering environments. Three broad problem areas inhibit such activities: (1) Computational, e.g, insufficient infrastructure for the sharing of very large amounts of information, results, and tools; (2) Structural institutional barriers, e.g., funding, publication, and promotion policies; and (3) Social, e.g., communication barriers stemming from narrow specialization. The participants supported specific steps to address these problems: explicit support and incentives for multidisciplinary activities; the development of digitial libraries to enhance interdisciplinary communication and understanding; and development of a “virtual scientific marketplace” to disseminate tools, results, and expertise

Electronic commerce in open large scale distributed environments presents several system level challenges in terms of execution model, infrastructure and management. The autonomy of providers severely restricts the scope of control by authorities other than the owner of each service, and makes the design of the distributed software architecture a challenging undertaking. We present our ongoing work toward the development of a distributed software substrate to support controlled interactions among components that are owned and managed by autonomous authorities. This infrastructure is based on a distributed component model, designed to support the paradigm of service level management for specifying and managing commitments made by independent service providers to their clients on performance related attributes of services.

This paper presents the Aurora architecture for network-centric applications, complementing the CORB A and WWW/Java frameworks with support for composition of services. The architecture addresses the requirements of dynamic open environments with multiple autonomous service providers. The paper presents a detailed technical design of the run-time infrastructure that enables on-demand composition of services. Aurora is based on a container framework, which provides the basis for dynamic and adaptive composition, as well as detailed monitoring and tracking. Work sessions are implemented as networks of active containers. A distinguishing feauture of Aurora is that such networks can be inspected and manipulated at run-time.

This paper focuses on technical issues regarding the challenge of integrating digital libraries in a work support environment that combines information management with coordination and collaboration. Such broad scope necessitates an extended view of electronic documents to allow for combinations of information content together with executable components. Moreover, as a result of the autonomy of the service and information content providers, this environment takes the form of an information economy. This paper provides a brief survey of research results related to architectures, metadata, interoperability and rights management in digital library systems. This survey serves to support a discussion of extending the scope of digital library systems toward integrated work support environments.

This paper discusses the requirements of current and emerging large-scale distributed applications and emphasizes the need for a common infrastructure to support them. A design for an infrastructure that aims at satisfying these requirements is presented. Moreover, it is shown how key aspects of important large-scale applications can exploit the services included in the proposed infrastructure. The paper concludes by discussing the current status of a prototype implementation and our research plan.

Περιέχει το πλήρες κείμενο Advances in information technology during the past decade have resulted in a proliferation of clinical information systems dedicated to di erent user groups and clinical functional areas. This, in turn, has created the need for hospital-wide management and integration of information, and has triggered major e orts towards the development of integrated hospital information systems as a building block of integrated regional healthcare networks. The diversity of hospital organiza- tions, the complexity of clinical protocols and proce- dures, as well as the di erent preferences of various user groups make it extremely di cult for a single monolithic information system to e ectively serve the needs of an entire healthcare organizational structure. Thus, infor- mation and telecommunications systems must primarily provide the infrastructure to support the e ective inte- gration of distributed and heterogeneous components, ensuring overall integrity in terms of functional and information interworking. This approach i.e., the inte- gration of heterogeneous autonomous distributed sys- tems, to developing and managing regional healthcare networks ensures the transfer and integration of consis- tent information between healthcare facilities, without imposing constraints on the operation of individual clinical units. This paper presents the results of an ongoing e ort for the design and implementation of an architecture based on digital library technologies, for the provision of user-oriented telematic services in a regional healthcare network. Speci®cally, it addresses issues related to the provision of user-oriented services, trans- parent to the needs of di erent user groups and the requirements of speci®c tasks, based on: a) meta-infor- mation for the creation of an information infrastructure for the regional healthcare network which is, e ectively, a multimedia distributed digital library, b) intelligent information retrieval strategies to selectively retrieve information from multimedia data, c) agent-based tech- nologies for e ective service delivery adapted to the current user needs and the task at hand, and d) middle ware services that explicitly reveal not only the charac- teristics of the information sources, but also address the context of speci®c telematic services, through appropriate mediation mechanisms.

This paper concerns the design of a system supporting teleconferencing applications. This system aims at creating an immersive environment by compositing a synthetic background with the reconstructed images of remote participants. It also presents natural scenes by employing viewer tracking and rendering images from the vantage point of the viewer. Processing requirements are met by exploiting existing networks of workstations by means of a quality of service based resource management scheme. This approach offers advantages with respect to the application itself (teleconferencing) and resource requirements. By restricting the transmitted images to the participants only, the required bandwidth becomes much smaller and video quality can be achieved. The employed resource management scheme limits computational resources to general purpose workstations. On the other hand, a synthetic background that matches the real meeting room can be easily designed, giving the impression of a virtual meeting room as an extension of the real one. In this way, remote conference sessions that resemble real life meetings can be held.

We consider commercial Digital Libraries as information economies consisting of several players: authors and publishers who create and sell their collections, suppliers (e.g. computer systems) who provide information storage, indexing and access services, information-agents who provide searching and presentation services, and users who request for services. In such an economic framework, one can envision suppliers and information-agents competing to provide services for information storage, searching, access and presentation. In providing such services, several issues arise; among them are socio-economic and cultural aspects of pricing information objects and Quality of Service (QoS) to access and view these objects. These issues play an important role in allocating resources such as processing time, network bandwidth and buffers, memory, cache and I/O, which are distributed (and owned) among various players in the economy. Using this framework, we present the interactions among the players, service models, pricing and charging/billing mechanisms, and corresponding implementation issues in large digital libraries. http://www.dlib.org/dlib/february96/forth/02sairamesh.html (1 de 10)02/01/2010 09:20:09 a.m. Economic Framework for Pricing and Charging In Digital Libraries

The paper discusses dynamic workload management in transaction processing systems where the workload consists of multiple classes of units of work, including workflows comprised of interdependent tasks. Business requirements specify that differing levels of service must be provided to different classes of work, thus it is natural to specify performance goals per work class, that reflect the business requirements for the work class as well as the inherent resource demands of the units of work. Adaptive algorithms have been proposed for the satisfaction of performance goals of transaction classes. Scheduling the execution of complete workflows, which are multi-transaction units of work, is complicated by the need for task coordination, due to both control and data flow dependencies among tasks. Current transaction processing monitors provide infrastructure for the coordination of tasks by means of queueing facilities. We draw on previous work on goal-oriented resource management to design adaptive task scheduling algorithms. A detailed simulator of transaction processing systems with a queueing facility has been developed, with the specific aim to study the performance for workloads that include multi-transaction units of work.

Managing the performance of distributed computing environments is today a serious challenge, because of their complexity and the diversity of new applications. Digital audio and video data, with strict delay and jitter constraints, are moved and manipulated in a computing, storage and communication medium where more traditional batch, transaction and query processing applications are already in operation and expected to achieve preset levels of throughput or average response time. Enterprise wide systems will soon surface where, for example, resources are shared among applications running high volume, low response time inventory control transactions with strong CPU time demands, applications managing videoconferences with tight delay and massive bandwidth requirements, engineering visualization tools with similar requirements, background decision support queries with soft completion deadlines and large memory needs, or process control tasks with hard real time constraints and demands for immediate use of CPU, memory or bandwidth.

Radiologists worldwide use mammography as a reliable tool for breast cancer screening. However, mammography assessment is challenging even for well-trained radiologists, leading to a pressing need for Computer Aided Diagnosis (CAD) systems. In this work, a novel technique for the detection and classification of breast Micro-Calcifications (MCs), which are diagnostically significant but difficult to detect findings, is presented. The proposed method is based on the subtraction of temporally sequential mammogram pairs, after pre-processing and image registration, followed by machine-learning. The classification was performed using several features extracted from the subtracted mammograms and selected during training to optimize the accuracy of the results. Six classifiers were tested in a leave-one-patient-out, 4, 5 and 10 fold cross-validation process. This technique was evaluated on a unique dataset, consisting of temporal sequences of mammograms from 80 patients taken between 1 to 6 years apart. The resulting 320 mammograms were reviewed by 2 radiologists who precisely marked each MC location. The accuracy of classifying MCs as benign or suspicious improved from 91.42% without temporal subtraction and an Ensemble of Decision Trees (EDT), to 99.55% with the use of sequential mammograms and Support Vector Machines (SVMs) with leave-one-patient-out validation. The improvement was statistically significant (p-value < 0.005). These results verify the accuracy and the effectiveness of the proposed technique should to be further evaluated on a larger dataset.

A flow control mechanism based on an economic model of optimum decentralized decision making is presented. The virtual circuits (VCs) of the network are considered as economic agents competing for link capacity resources. Each VC is equipped with a preference relation defining its individual throughput-delay goals. The work makes three contributions to the study of flow control protocols. First, it presents fully decentralized flow control algorithms that allocate capacity to reflect both, varying priorities among VCs as well as their preferences and needs. Second, it establishes that the flow control algorithms result in Pareto-optimal allocation of capacities. This implies that the algorithms accomplish optimally fair allocation of resources given the set priorities and preferences. Third, experimental evidence is provided that the algorithms converge rapidly. >