A Gradient based algorithm for blind inversion of Wiener System using multi-variate score functions

A Wiener system is a linear time-invariant filter, followed by an invertible nonlinear distortion. Assuming that the input signal is an independent and identically distributed (iid) sequence, we propose an algorithm for estimating the input signal only by observing the output of the Wiener system. The algorithm is based on minimizing the mutual information of the output samples, by means of a steepest descent gradient approach.

Improving algorithm speed in PNL mixture separation and Wiener system inversion

This paper proposes a very simple method for increasing the algorithm speed for separating sources from PNL mixtures or invertingWiener systems. The method is based on a pertinent initialization of the inverse system, whose computational cost is very low. The nonlinear part is roughly approximated by pushing the observations to be Gaussian; this method provides a surprisingly good approximation even when the basic assumption is not fully satisfied. The linear part is initialized so that outputs are decorrelated. Experiments shows the impressive speed improvement.

Nonlinear prediction based on score function

The linear prediction coding of speech is based in the assumption that the generation model is autoregresive. In this paper we propose a structure to cope with the nonlinear effects presents in the generation of the speech signal. This structure will consist of two stages, the first one will be a classical linear prediction filter, and the second one will model the residual signal by means of two nonlinearities between a linear filter. The coefficients of this filter are computed by means of a gradient search on the score function. This is done in order to deal with the fact that the probability distribution of the residual signal still is not gaussian. This fact is taken into account when the coefficients are computed by a ML estimate. The algorithm based on the minimization of a high-order statistics criterion, uses on-line estimation of the residue statistics and is based on blind deconvolution of Wiener systems [1]. Improvements in the experimental results with speech signals emphasize on the interest of this approach.

An efficient total variation algorithm for super-resolution in fetal brain MRI with adaptive regularization.

Although fetal anatomy can be adequately viewed in new multi-slice MR images, many critical limitations remain for quantitative data analysis. To this end, several research groups have recently developed advanced image processing methods, often denoted by super-resolution (SR) techniques, to reconstruct from a set of clinical low-resolution (LR) images, a high-resolution (HR) motion-free volume. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has been quite attracted by Total Variation energies because of their ability in edge preserving but only standard explicit steepest gradient techniques have been applied for optimization. In a preliminary work, it has been shown that novel fast convex optimization techniques could be successfully applied to design an efficient Total Variation optimization algorithm for the super-resolution problem. In this work, two major contributions are presented. Firstly, we will briefly review the Bayesian and Variational dual formulations of current state-of-the-art methods dedicated to fetal MRI reconstruction. Secondly, we present an extensive quantitative evaluation of our SR algorithm previously introduced on both simulated fetal and real clinical data (with both normal and pathological subjects). Specifically, we study the robustness of regularization terms in front of residual registration errors and we also present a novel strategy for automatically select the weight of the regularization as regards the data fidelity term. Our results show that our TV implementation is highly robust in front of motion artifacts and that it offers the best trade-off between speed and accuracy for fetal MRI recovery as in comparison with state-of-the art methods.

New Algorithm for Managing Childhood Illness Using Mobile Technology (ALMANACH): A Controlled Non-Inferiority Study on Clinical Outcome and Antibiotic Use in Tanzania.

INTRODUCTION: The decline of malaria and scale-up of rapid diagnostic tests calls for a revision of IMCI. A new algorithm (ALMANACH) running on mobile technology was developed based on the latest evidence. The objective was to ensure that ALMANACH was safe, while keeping a low rate of antibiotic prescription. METHODS: Consecutive children aged 2-59 months with acute illness were managed using ALMANACH (2 intervention facilities), or standard practice (2 control facilities) in Tanzania. Primary outcomes were proportion of children cured at day 7 and who received antibiotics on day 0. RESULTS: 130/842 (15∙4%) in ALMANACH and 241/623 (38∙7%) in control arm were diagnosed with an infection in need for antibiotic, while 3∙8% and 9∙6% had malaria. 815/838 (97∙3%;96∙1-98.4%) were cured at D7 using ALMANACH versus 573/623 (92∙0%;89∙8-94∙1%) using standard practice (p<0∙001). Of 23 children not cured at D7 using ALMANACH, 44% had skin problems, 30% pneumonia, 26% upper respiratory infection and 13% likely viral infection at D0. Secondary hospitalization occurred for one child using ALMANACH and one who eventually died using standard practice. At D0, antibiotics were prescribed to 15∙4% (12∙9-17∙9%) using ALMANACH versus 84∙3% (81∙4-87∙1%) using standard practice (p<0∙001). 2∙3% (1∙3-3.3) versus 3∙2% (1∙8-4∙6%) received an antibiotic secondarily. CONCLUSION: Management of children using ALMANACH improve clinical outcome and reduce antibiotic prescription by 80%. This was achieved through more accurate diagnoses and hence better identification of children in need of antibiotic treatment or not. The building on mobile technology allows easy access and rapid update of the decision chart. TRIAL REGISTRATION: Pan African Clinical Trials Registry PACTR201011000262218.

Managing the Sick Child in the Era of Declining Malaria Transmission: Development of ALMANACH, an Electronic Algorithm for Appropriate Use of Antimicrobials.

OBJECTIVE: To review the available knowledge on epidemiology and diagnoses of acute infections in children aged 2 to 59 months in primary care setting and develop an electronic algorithm for the Integrated Management of Childhood Illness to reach optimal clinical outcome and rational use of medicines. METHODS: A structured literature review in Medline, Embase and the Cochrane Database of Systematic Review (CDRS) looked for available estimations of diseases prevalence in outpatients aged 2-59 months, and for available evidence on i) accuracy of clinical predictors, and ii) performance of point-of-care tests for targeted diseases. A new algorithm for the management of childhood illness (ALMANACH) was designed based on evidence retrieved and results of a study on etiologies of fever in Tanzanian children outpatients. FINDINGS: The major changes in ALMANACH compared to IMCI (2008 version) are the following: i) assessment of 10 danger signs, ii) classification of non-severe children into febrile and non-febrile illness, the latter receiving no antibiotics, iii) classification of pneumonia based on a respiratory rate threshold of 50 assessed twice for febrile children 12-59 months; iv) malaria rapid diagnostic test performed for all febrile children. In the absence of identified source of fever at the end of the assessment, v) urine dipstick performed for febrile children <2 years to consider urinary tract infection, vi) classification of 'possible typhoid' for febrile children >2 years with abdominal tenderness; and lastly vii) classification of 'likely viral infection' in case of negative results. CONCLUSION: This smartphone-run algorithm based on new evidence and two point-of-care tests should improve the quality of care of <5 year children and lead to more rational use of antimicrobials.

Eye-tracking of nodule detection in lung CT volumetric data.

PURPOSE: Signal detection on 3D medical images depends on many factors, such as foveal and peripheral vision, the type of signal, and background complexity, and the speed at which the frames are displayed. In this paper, the authors focus on the speed with which radiologists and naïve observers search through medical images. Prior to the study, the authors asked the radiologists to estimate the speed at which they scrolled through CT sets. They gave a subjective estimate of 5 frames per second (fps). The aim of this paper is to measure and analyze the speed with which humans scroll through image stacks, showing a method to visually display the behavior of observers as the search is made as well as measuring the accuracy of the decisions. This information will be useful in the development of model observers, mathematical algorithms that can be used to evaluate diagnostic imaging systems. METHODS: The authors performed a series of 3D 4-alternative forced-choice lung nodule detection tasks on volumetric stacks of chest CT images iteratively reconstructed in lung algorithm. The strategy used by three radiologists and three naïve observers was assessed using an eye-tracker in order to establish where their gaze was fixed during the experiment and to verify that when a decision was made, a correct answer was not due only to chance. In a first set of experiments, the observers were restricted to read the images at three fixed speeds of image scrolling and were allowed to see each alternative once. In the second set of experiments, the subjects were allowed to scroll through the image stacks at will with no time or gaze limits. In both static-speed and free-scrolling conditions, the four image stacks were displayed simultaneously. All trials were shown at two different image contrasts. RESULTS: The authors were able to determine a histogram of scrolling speeds in frames per second. The scrolling speed of the naïve observers and the radiologists at the moment the signal was detected was measured at 25-30 fps. For the task chosen, the performance of the observers was not affected by the contrast or experience of the observer. However, the naïve observers exhibited a different pattern of scrolling than the radiologists, which included a tendency toward higher number of direction changes and number of slices viewed. CONCLUSIONS: The authors have determined a distribution of speeds for volumetric detection tasks. The speed at detection was higher than that subjectively estimated by the radiologists before the experiment. The speed information that was measured will be useful in the development of 3D model observers, especially anthropomorphic model observers which try to mimic human behavior.

Search Interfaces on the Web: Querying and Characterizing

Current-day web search engines (e.g., Google) do not crawl and index a significant portion of theWeb and, hence, web users relying on search engines only are unable to discover and access a large amount of information from the non-indexable part of the Web. Specifically, dynamic pages generated based on parameters provided by a user via web search forms (or search interfaces) are not indexed by search engines and cannot be found in searchers’ results. Such search interfaces provide web users with an online access to myriads of databases on the Web. In order to obtain some information from a web database of interest, a user issues his/her query by specifying query terms in a search form and receives the query results, a set of dynamic pages that embed required information from a database. At the same time, issuing a query via an arbitrary search interface is an extremely complex task for any kind of automatic agents including web crawlers, which, at least up to the present day, do not even attempt to pass through web forms on a large scale. In this thesis, our primary and key object of study is a huge portion of the Web (hereafter referred as the deep Web) hidden behind web search interfaces. We concentrate on three classes of problems around the deep Web: characterization of deep Web, finding and classifying deep web resources, and querying web databases. Characterizing deep Web: Though the term deep Web was coined in 2000, which is sufficiently long ago for any web-related concept/technology, we still do not know many important characteristics of the deep Web. Another matter of concern is that surveys of the deep Web existing so far are predominantly based on study of deep web sites in English. One can then expect that findings from these surveys may be biased, especially owing to a steady increase in non-English web content. In this way, surveying of national segments of the deep Web is of interest not only to national communities but to the whole web community as well. In this thesis, we propose two new methods for estimating the main parameters of deep Web. We use the suggested methods to estimate the scale of one specific national segment of the Web and report our findings. We also build and make publicly available a dataset describing more than 200 web databases from the national segment of the Web. Finding deep web resources: The deep Web has been growing at a very fast pace. It has been estimated that there are hundred thousands of deep web sites. Due to the huge volume of information in the deep Web, there has been a significant interest to approaches that allow users and computer applications to leverage this information. Most approaches assumed that search interfaces to web databases of interest are already discovered and known to query systems. However, such assumptions do not hold true mostly because of the large scale of the deep Web – indeed, for any given domain of interest there are too many web databases with relevant content. Thus, the ability to locate search interfaces to web databases becomes a key requirement for any application accessing the deep Web. In this thesis, we describe the architecture of the I-Crawler, a system for finding and classifying search interfaces. Specifically, the I-Crawler is intentionally designed to be used in deepWeb characterization studies and for constructing directories of deep web resources. Unlike almost all other approaches to the deep Web existing so far, the I-Crawler is able to recognize and analyze JavaScript-rich and non-HTML searchable forms. Querying web databases: Retrieving information by filling out web search forms is a typical task for a web user. This is all the more so as interfaces of conventional search engines are also web forms. At present, a user needs to manually provide input values to search interfaces and then extract required data from the pages with results. The manual filling out forms is not feasible and cumbersome in cases of complex queries but such kind of queries are essential for many web searches especially in the area of e-commerce. In this way, the automation of querying and retrieving data behind search interfaces is desirable and essential for such tasks as building domain-independent deep web crawlers and automated web agents, searching for domain-specific information (vertical search engines), and for extraction and integration of information from various deep web resources. We present a data model for representing search interfaces and discuss techniques for extracting field labels, client-side scripts and structured data from HTML pages. We also describe a representation of result pages and discuss how to extract and store results of form queries. Besides, we present a user-friendly and expressive form query language that allows one to retrieve information behind search interfaces and extract useful data from the result pages based on specified conditions. We implement a prototype system for querying web databases and describe its architecture and components design.

Development of phase retrieval algorithm and techniques for optical spectrum analysis

Coherent anti-Stokes Raman scattering is the powerful method of laser spectroscopy in which significant successes are achieved. However, the non-linear nature of CARS complicates the analysis of the received spectra. The objective of this Thesis is to develop a new phase retrieval algorithm for CARS. It utilizes the maximum entropy method and the new wavelet approach for spectroscopic background correction of a phase function. The method was developed to be easily automated and used on a large number of spectra of different substances.. The algorithm was successfully tested on experimental data.

Efficient total variation algorithm for fetal MRI reconstruction

Fetal MRI reconstruction aims at finding a high-resolution image given a small set of low-resolution images. It is usually modeled as an inverse problem where the regularization term plays a central role in the reconstruction quality. Literature has considered several regularization terms s.a. Dirichlet/Laplacian energy [1], Total Variation (TV)based energies [2,3] and more recently non-local means [4]. Although TV energies are quite attractive because of their ability in edge preservation, standard explicit steepest gradient techniques have been applied to optimize fetal-based TV energies. The main contribution of this work lies in the introduction of a well-posed TV algorithm from the point of view of convex optimization. Specifically, our proposed TV optimization algorithm for fetal reconstruction is optimal w.r.t. the asymptotic and iterative convergence speeds O(1/n(2)) and O(1/root epsilon), while existing techniques are in O(1/n) and O(1/epsilon). We apply our algorithm to (1) clinical newborn data, considered as ground truth, and (2) clinical fetal acquisitions. Our algorithm compares favorably with the literature in terms of speed and accuracy.

Search and Discovery : OER's Open Loop

Open educational resources (OER) promise increased access, participation, quality, and relevance, in addition to cost reduction. These seemingly fantastic promises are based on the supposition that educators and learners will discover existing resources, improve them, and share the results, resulting in a virtuous cycle of improvement and re-use. By anecdotal metrics, existing web scale search is not working for OER. This situation impairs the cycle underlying the promise of OER, endangering long term growth and sustainability. While the scope of the problem is vast, targeted improvements in areas of curation, indexing, and data exchange can improve the situation, and create opportunities for further scale. I explore the way the system is currently inadequate, discuss areas for targeted improvement, and describe a prototype system built to test these ideas. I conclude with suggestions for further exploration and development.