Search

Image segmentation is a fundamental task for automatic disease diagnostics, and UNET is a widely used architecture for medical image segmentation. This is because this architecture is relatively simple to implement and obtains robust results. Currently there are several databases of melanocytic lesions, so determining which databases and the amount of data needed to train on such an architecture is a challenge. Another drawback is often to generate a proper database, since the ideal segmentation needed for training requires a large effort on the part of medical professionals. In this paper, we analyze the impact of melanocytic lesion databases on pretraining using the UNET architecture and its impact on segmentation results. ResNet with pre-trained weights with Imagenet is used. Three different databases were used in the experiments. Experimental results show that models pre-trained with images of melanocytic lesions perform better than those models that do not use pre-training for segmentation. In turn, adding more models pre-trained with melanocytic lesion database gives better results in most cases. Finally, for most of the tests performed it can be observed that using Imagenet only in pre-training obtains worse results than combining pretraining with other melanocytic lesion databases.

This work investigated and implemented the integration and effective use of Computer Aided Diagnostic and Detection systems (CAD) with a Picture Archiving and Communication System (PACS) and how this could provide standardized data and expand the coverage of specialized professionals. Likewise, standardized data allowed interoperability between systems. The designed CAD is focused on Diabetic Retinopathy diagnosis with three main functions: (I) digital image processing for detection, segmentation and extraction of retina image anomaly features, (II) binary classification, and finally (III) converter of the results to the DICOM standard (Digital Imagine and Communication Systems). It was decided to use an open source and free license PACS called Orthanc. The results show the successful integration of the systems, since the studies entered in the CAD are accessed directly from the PACS. At the same time, usability tests show a high degree of efficiency, effectiveness and satisfaction of clinical users in the use of the system.

This work proposes a solution to the problem of management automation in the Emergency Department of the Hospital de Clinicas of the National University of Asuncion. This tool will generate benefits for both the patient and the hospital's target staff, since by computerizing the processes, which is still done manually, there will be more time to dedicate better care to the patient and doctors could spend part of their time analyzing the statistics and information, that can be generated through the application to carry out investigations in the emergency care area, as well as optimize the resources of both, the staff and the logistics used in the Hospital. To achieve the main objective of this work, an application based on international health standards has been developed, as well as good practices in the care processes in an extremely sensitive area such as the Emergency Department. With this application it is possible to innovate in the Emergency Department, given that after more than 100 years of history of the Hospital, for the first time it intends to automate processes and generate online information quickly and efficiently from this modern tool optimizing attention to patients in the emergency area.

The huge bandwidth exploited in optical fibers and the ability to handle multiple simultaneous transmissions on the same fiber due to the WDM technology, have made the problem of protection, multicast routing and wavelength allocation (MPRWA) critical for the success of point-to-multipoint applications. In order to maintain the quality of service required by these applications, the network faces the restriction of rapid recovery in cases of failure. Also, it must minimize the different costs that this entails, and prioritize requests in case of not having the necessary resources for recovery. In this context, this work deals with the design of the main multicast route and its protection, with quality of protection (QoP) levels. For this reason, two protection schemes have been addressed: dualtree based and biconnected-subgraph based. To achieve this, competitive evolutionary techniques are applied; where the total number of links used, the number of wavelength converters, the number of splitter nodes, and the number of destinations served and protected are objective functions simultaneously optimized in a Pareto context. The experimental tests were carried out on different network topologies and multicast demands, considering the hypervolume as a Pareto quality measure. The results suggest that the subgraph-based strategy is more promising, obtaining better results than the protection based on dual-tree.

Medical images are used to analyze and diagnose diseases. The analysis of these diseases requires images with high contrast and detail, because they contain important information about the disease. Capture methods are improving every day, but sometimes the quality of medical images is not adequate to make a good medical diagnosis. Some common problems are ambient noise, poor lighting conditions, or technical limitations of the devices generating low quality medical images. For these cases, contrast enhancement techniques, which enhance the visual quality of the images, are generally useful. However, there is still room for improvement. In this paper, we propose a novel algorithm to improve the contrast and detail of medical images. This proposal is based on a multiscale morphological approach that uses two structuring elements in the top-hat transform. The simulations were performed with medical images taken from public repositories. The experimental results show that the proposal enhances local contrast and improves the details of the medical images.

Software defined networking (SDN) is an emerging technology based on the separation of the control plane and the data plane. This allows to obtain benefits, in comparison with traditional networks, in terms of network management, global monitoring-control, cost reduction, and in particular the energy saving by the strategic activation of devices. In this paper, we propose an approach that seeks to minimize the global energy consumption of the network by suspending inactive devices, such as chassis and line cards, as well as limiting the use of links in traffic sessions. For this purpose, we developed an Integer Linear Programming (ILP) model for the SDN routing problem in order to obtain the minimum energy consumption, subject to satisfy all traffic demands. The experimental results on two network topologies for a set of static traffic requests indicate that the proposed model is promising, saving up to 42% of the global energy consumption obtaining a better performance to the models proposed in the literature. On the other hand, the experimental results for incremental semi-dynamic traffic indicate that the performance of the optimization with re-routing improves the approach without re-routing when increasing the traffic in the network, but this improvement is not always perceptible. The approach without re-routing in terms of scalability is promising, by increasing the traffic load not generate interruptions to the traffic already attended and affect the quality of the service.

TapeYty is a tool developed to calculate optimal paths for urban garbage collection vehicles in the Asuncioon city. This tool generates benefits mainly in the economic and environmental aspects of the city where a large number of people brings a high generation of waste. This makes the complexity of garbage management even greater. The routing problem is treated as the open rural postman problem which seeks to minimize the distance to be traveled by the collection vehicles. To achieve the objective TapeYty is based on mathematical programming techniques and Geographical Informatiom System (GIS) tool which allows the management of the route network, being able able to update the road way and their state of blocked or nonblocked. This implies that when changes of state of the streets TapeYty modifies the graph that represents the road network and re-calculates the solutions providing new optimal routes to each vehicle of collection. The tool has provided solutions that save on average 20% of distance traveled compared to current tour.

This paper proposes a solution to the problem of automated management of Acute Lymphoblastic Leukemia (ALL) in children that are being treated at the Hospital de Clinicas (HC) which belongs to the Faculty of Medical Sciences of the National University of Asuncion (FCM-UNA). For this purpose, we are proposing the design and implementation of a computer application called Pediatric Hemato-Oncological Protocols Automation (APHOPe), which facilitates obtaining information and managing the implementation of the procedure protocols and hematological and oncological treatment used in the Pediatric Hemato-Oncology Department (HOPe). APHOPe can provide statistical data and applied treatments so the effectiveness-efficiency relation of the global and particular protocols for each patient can be analyzed. In this context, APHOPe allows the generation of various Ad Hoc protocols depending on the treatment needs of the medical personnel. In this stage the system provides the 2008 Protocol LLA which is a variant of the Berlin-Frankfurt-Munster Protocol. The usability and acceptability tests indicate that the proposed system is an effective and flexible tool for the automated management of the hemato-oncological protocols.

Searching for adequate input parameters of a Contrast Enhancement Algorithm is a fundamental task, aimed to get more suitable images, in terms of contrast, for decision making or further applications. Several Optimization approaches take a single image as the input of the process, so they get poor performance. In this proposal, a group of several images of the same type are taken as input of a Robust Multi-Objective Particle Swarm Optimization approach, in order to achieve more suitable input parameters for a kind of images, instead of a single image. The results are a set of input parameters for the Contrast Enhancement Algorithm, suitable for Contrast Enhancement of a group of images of the same type.

Fragmentation in Elastic Optical Networks is an issue caused by isolated, non-aligned, and non-contiguous frequency slots that can not be used to allocate new connection request to the network, due to the optical layer restrictions imposed to the Routing and Spectrum Assignment (RSA) algorithms. To deal with this issue, several studies about Spectrum Defragmentation have been presented. In this work, we analyze the most important Non-Hitless Defragmentation Algorithms found in the literature, with proactive and reactive approaches that include rerouting and non-rerouting schemes, and compare their performance in terms of Blocking Probability, Entropy, and Bandwidth Fragmentation Ratio. Simulations results showed that the Fragmentation Aware schemes outperformed the other algorithms in low traffic load, but the Reactive schemes got better results in high traffic load.

Estimating effort is a very important task in any organization. Significant over or under-estimates can be very expensive for software project companies. The use of computing intelligence methods has been recently proposed for software development effort estimation. In this study, we present new models to estimate the effort required for the development of software projects. These new models were calculated using Linear Genetic Programming (PGL). The results show that the proposed models get more precise and more effective estimation for Mean Magnitude Relative error (MMRE) and Mean Magnitude of Relative Error relative to the Estimate (MMER) than using the constructive cost model (COCOMO). We performed the study based on three stages according to the type of project. The models were designed and validated by simulation with the public repository dataset COCOMO81 and NASA93. Performance of the proposed models EE_PGLa, EE_PGLb and EE_PGLc are more accurate than COCOMO.

This paper presents the super symmetry model of permutations groups. It is inspired by super symmetry, in which theoretical physicists postulate the existence of super partners for each fundamental particle of the standard model. Under these same ideas and analyzes on the structures of permutations, one-one correspondence (or entanglements) was found in the set of permutations. In this work is proposed exploit the benefits of symmetry to split the search space of combinatorial optimization problems on permutations. Therefore, a strategy exploration landscape objective function was developed based on the entanglement, which is simultaneously evaluate related permutations. The experimental results indicate that this approach provides improved computational time brute-force search and higher quality solutions on genetic algorithms is achieved.

The higth demand of bandwidth by the diversification of the technologies in Internet have promoted the development of a new paradigm for the optical networks that is known as Elastics Optical Network (EON) which changes the fixed grid of the spectrum in WDM networks by a flexible grid approach to use more efficiently the available spectrum. One of the principal challenges that exists to implement EON is to find a algorithm of routing and spectrum assignment (RSA) to serve the incoming demands. In this work is proposed EONS, a simulator where can be emulated the execution of Dinamics RSA algorithms and shows the blocking probabilities to comparison of the efficiency of these algorithms. For this work, basic RSA algorithms were implemented to verify the correct operation of EONS in different networks and traffic loads.

Given the importance and complexity of the problem of robust network design, this work, studies the robust network design subject to guarantee certain level of quality of service, so that the reservation of an adjustable bandwidth for each node, the network is not negatively influenced by traffic from the rest of the network. Therefore a MultiObjective Evolutionary Algorithm (MOEA) is proposed to solve and find the robust network design, inspired by the concept of price of robustness, which simultaneously minimizes the cost of network, inequity traffic and maximizes the traffic service in the worst case scenario. Finally, experimental results show the benefits of the proposed approach to get a set of non-dominated solution on which it can make, a better decision making.

Melanoma is a type of skin cancer and is caused by the uncontrolled growth of atypical melanocytes. In recent decades, computer aided diagnosis is used to support medical professionals; however, there is still no globally accepted tool. In this context, similar to state-of-the-art we propose a system that receives a dermatoscopy image and provides a diagnostic if the lesion is benign or malignant. This tool is based on next modules: Preprocessing, Segmentation, Feature Extraction and Classification. Preprocessing involves the removal of hairs. Segmentation is to isolate the lesion. Feature extraction is considering the ABCD dermoscopy rule. The classification is performed by the Support Vector Machine. Experimental evidence indicates that the proposal has 90.63 % accuracy, 95 % sensitivity and 83.33 % specificity on a dataset of 104 dermatoscopy images. These results are favorable considering the performance of diagnosis by traditional progress in the area of dermatology.

In certain medical images, it is possible to achieve contrast enhancement at different levels, in order to highlight different structures present therein. This could be useful to medical specialists to perform more specific diagnoses, in chest radiographs and mammograms, where it is possible to highlight different details when contrast is enhanced. Parameter tuning for Contrast Limited Adaptive Histogram Equalization (CLAHE) using a multi-objective meta-heuristic (SMPSO) is proposed, where the objective functions are the maximization of the amount of information available (via Entropy) and minimization of distortion in the resulting images (Structural Similarity Index, SSIM) simultaneously. The results show that our approach calculates a set of non-dominated solutions or Pareto Set, which represents images with different contrast levels and different levels of commitment between Entropy and Structural Similarity Index. Particularly, these objective functions are contradictory. These enhanced images provide useful information for decision making of specialists.

This work deals with the survivable optical network design problem for static multicast traffic, subject to simple node fails. Given a set of multicast request, it is proposed an algorithm based on Multi-Ojective Ant Colony Optimization which tries to find the best network design, as well as the primary and back-up multicast trees, with protection against node fails. The proposed algorithm simultaneously minimizes the network design cost and the maximum end-to-end optical delay. The experimental results over different instances show the benefits of three protection approaches based on total or partial network reconfiguration.

A network design is robust if it is able to deal with any traffic requirement under certain bounds and physical network conditions. The robust network design is a complex problem of growing importance where, in general, the only information available are traffic bounds of the network links. This work proposes a Genetic Algorithm to design robust networks with optimal capacity of links considering a stable routing with uncertain traffic that can be divided in k sub-routes. The uncertain traffic is handled by usign the hose model which imposes a maximum input/output traffic for each network node. Experimental results for a set of instances with different number of routes show the convenience of a stable routing against a non-divisible routing (k = 1). However, an increasing of the k value implies an increasing of the number of viable solutions, thus, a trade-off relation between k and the quality of solutions obtained by the proposed algorithm was detected.

Wavelength Division Multiplexing (WDM) networks are designed for long-term operations in which the uncertainty of future traffic plays an important role. The performance of these networks is highly dependent on the routing and the wavelength-converter allocation algorithms. In order to achieve a good performance on a long-term basis, both problems have to be dealt with together and traffic uncertainty should be included. To this aim, this work proposes a joint optimization approach where the converters allocation plan and paths for routing are calculated simultaneously. A Multi-Objective Evolutionary Algorithm (MOEA) is proposed to minimize the number of wavelength converters, the average blocking probability, and an unfairness measure, all three at the same time, the uncertainty is modelled by means of a set of scenarios. The proposed MOEA calculates an approximation to the optimal set of solutions. This work presents experimental results showing the feasibility of the proposed approach in a multi-objective and uncertain context.