A Food Recognition System for Diabetic Patients based on an Optimized Bag of Features Model

Computer vision-based food recognition could be used to estimate a meal's carbohydrate content for diabetic patients. This study proposes a methodology for automatic food recognition, based on the Bag of Features (BoF) model. An extensive technical investigation was conducted for the identification and optimization of the best performing components involved in the BoF architecture, as well as the estimation of the corresponding parameters. For the design and evaluation of the prototype system, a visual dataset with nearly 5,000 food images was created and organized into 11 classes. The optimized system computes dense local features, using the scale-invariant feature transform on the HSV color space, builds a visual dictionary of 10,000 visual words by using the hierarchical k-means clustering and finally classifies the food images with a linear support vector machine classifier. The system achieved classification accuracy of the order of 78%, thus proving the feasibility of the proposed approach in a very challenging image dataset.

Lateralized processing of faces: The role of features, configurations and familiarity

We investigated the lateralized processing of featural and configural information in face recognition in two divided visual field studies. In Experiment 1, participants matched the identity of a cue face containing either featural (scrambled faces) or configural (blurred faces) information with an intact test face presented subsequently either in the right visual field (RVF) or in the left visual field (LVF). Unilateral presentation was controlled by monitoring eye movements. The results show an advantage of the left hemisphere (LH) over the right hemisphere (RH) for featural processing and a specialization of the RH for configural compared to featural processing. In Experiment 2, we focused on configural processing and its relationship to familiarity. Either learned or novel test faces were presented in the LVF or the RVF. Participants recognized learned faces better when presented in the LVF than in the RVF, suggesting that the RH has an advantage in the recognition of learned faces. Because the recognition of familiar faces relies strongly on configural information (Buttle & Raymond, 2003), we argue that the advantage of the RH over the LH in configural processing is a function of familiarity.

Phenotypes of childhood asthma: are they real?

It has been suggested that there are several distinct phenotypes of childhood asthma or childhood wheezing. Here, we review the research relating to these phenotypes, with a focus on the methods used to define and validate them. Childhood wheezing disorders manifest themselves in a range of observable (phenotypic) features such as lung function, bronchial responsiveness, atopy and a highly variable time course (prognosis). The underlying causes are not sufficiently understood to define disease entities based on aetiology. Nevertheless, there is a need for a classification that would (i) facilitate research into aetiology and pathophysiology, (ii) allow targeted treatment and preventive measures and (iii) improve the prediction of long-term outcome. Classical attempts to define phenotypes have been one-dimensional, relying on few or single features such as triggers (exclusive viral wheeze vs. multiple trigger wheeze) or time course (early transient wheeze, persistent and late onset wheeze). These definitions are simple but essentially subjective. Recently, a multi-dimensional approach has been adopted. This approach is based on a wide range of features and relies on multivariate methods such as cluster or latent class analysis. Phenotypes identified in this manner are more complex but arguably more objective. Although phenotypes have an undisputed standing in current research on childhood asthma and wheezing, there is confusion about the meaning of the term 'phenotype' causing much circular debate. If phenotypes are meant to represent 'real' underlying disease entities rather than superficial features, there is a need for validation and harmonization of definitions. The multi-dimensional approach allows validation by replication across different populations and may contribute to a more reliable classification of childhood wheezing disorders and to improved precision of research relying on phenotype recognition, particularly in genetics. Ultimately, the underlying pathophysiology and aetiology will need to be understood to properly characterize the diseases causing recurrent wheeze in children.

Description of an ancient social bee trapped in amber using diagnostic radioentomology

The application of non-invasive imaging technologies using X-radiation (diagnostic radioentomology, ‘DR’) is demonstrated for the study of amber-entombed social bees. Here, we examine the external and internal morphology of an Early Miocene (Burdigalian) stingless bee (Apinae: Meliponini) from the Dominican Republic using non-destructive X-ray microtomography analysis. The study permits the accurate reconstruction of features otherwise obscured or impossible to visualize without destroying the sample and allows diagnosis of the specimen as a new species, Proplebeia adbita Greco and Engel.

Modeling Features at Runtime

A feature represents a functional requirement fulfilled by a system. Since many maintenance tasks are expressed in terms of features, it is important to establish the correspondence between a feature and its implementation in source code. Traditional approaches to establish this correspondence exercise features to generate a trace of runtime events, which is then processed by post-mortem analysis. These approaches typically generate large amounts of data to analyze. Due to their static nature, these approaches do not support incremental and interactive analysis of features. We propose a radically different approach called live feature analysis, which provides a model at runtime of features. Our approach analyzes features on a running system and also makes it possible to grow feature representations by exercising different scenarios of the same feature, and identifies execution elements even to the sub-method level. We describe how live feature analysis is implemented effectively by annotating structural representations of code based on abstract syntax trees. We illustrate our live analysis with a case study where we achieve a more complete feature representation by exercising and merging variants of feature behavior and demonstrate the efficiency or our technique with benchmarks.

New method for quantification and statistical analysis of nociceptive reflex receptive fields in humans

A method for quantifying nociceptive withdrawal reflex receptive fields in human volunteers and patients is described. The reflex receptive field (RRF) for a specific muscle denotes the cutaneous area from which a muscle contraction can be evoked by a nociceptive stimulus. The method is based on random stimulations presented in a blinded sequence to 10 stimulation sites. The sensitivity map is derived by interpolating the reflex responses evoked from the 10 sites. A set of features describing the size and location of the RRF is presented based on statistical analysis of the sensitivity map within every subject. The features include RRF area, volume, peak location and center of gravity. The method was applied to 30 healthy volunteers. Electrical stimuli were applied to the sole of the foot evoking reflexes in the ankle flexor tibialis anterior. The RRF area covered a fraction of 0.57+/-0.06 (S.E.M.) of the foot and was located on the medial, distal part of the sole of the foot. An intramuscular injection into flexor digitorum brevis of capsaicin was performed in one spinal cord injured subject to attempt modulation of the reflex receptive field. The RRF area, RRF volume and location of the peak reflex response appear to be the most sensitive measures for detecting modulation of spinal nociceptive processing. This new method has important potential applications for exploring aspects of central plasticity in volunteers and patients. It may be utilized as a new diagnostic tool for central hypersensitivity and quantification of therapeutic interventions.

Computer-aided diagnosis of carotid atherosclerosis based on ultrasound image statistics, laws' texture and neural networks

Quantitative characterisation of carotid atherosclerosis and classification into symptomatic or asymptomatic is crucial in planning optimal treatment of atheromatous plaque. The computer-aided diagnosis (CAD) system described in this paper can analyse ultrasound (US) images of carotid artery and classify them into symptomatic or asymptomatic based on their echogenicity characteristics. The CAD system consists of three modules: a) the feature extraction module, where first-order statistical (FOS) features and Laws' texture energy can be estimated, b) the dimensionality reduction module, where the number of features can be reduced using analysis of variance (ANOVA), and c) the classifier module consisting of a neural network (NN) trained by a novel hybrid method based on genetic algorithms (GAs) along with the back propagation algorithm. The hybrid method is able to select the most robust features, to adjust automatically the NN architecture and to optimise the classification performance. The performance is measured by the accuracy, sensitivity, specificity and the area under the receiver-operating characteristic (ROC) curve. The CAD design and development is based on images from 54 symptomatic and 54 asymptomatic plaques. This study demonstrates the ability of a CAD system based on US image analysis and a hybrid trained NN to identify atheromatous plaques at high risk of stroke.

Bathymetry of Lake Constance – a High-Resolution Survey in a Large, Deep Lake

Within the project “Tiefenschärfe – hochauflösende Vermes- sung Bodensee” a high-resolution seamless terrain model is created using airborne topobathymetric laserscanning and multibeam echosounder (MBES) techniques. The project visu- alizes the enormous wealth of features of underwater land- scapes of lakes. The combination of hydroacoustic (multibeam echosounder) and laser-optic (topobathymetric laserscan- ning) methods was used for the first time in a freshwater body of this size. Opportunities, limitations and restrictions of these high-resolution methods are presented.

Towards Automatic Identification of Discourse Markers in Dialogs: the Case of 'Like'

This article discusses the detection of discourse markers (DM) in dialog transcriptions, by human annotators and by automated means. After a theoretical discussion of the definition of DMs and their relevance to natural language processing, we focus on the role of like as a DM. Results from experiments with human annotators show that detection of DMs is a difficult but reliable task, which requires prosodic information from soundtracks. Then, several types of features are defined for automatic disambiguation of like: collocations, part-of-speech tags and duration-based features. Decision-tree learning shows that for like, nearly 70% precision can be reached, with near 100% recall, mainly using collocation filters. Similar results hold for well, with about 91% precision at 100% recall.

CompAS: A new approach to commonality and variability analysis with applications in computer assisted orthopaedic surgery

In rapidly evolving domains such as Computer Assisted Orthopaedic Surgery (CAOS) emphasis is often put first on innovation and new functionality, rather than in developing the common infrastructure needed to support integration and reuse of these innovations. In fact, developing such an infrastructure is often considered to be a high-risk venture given the volatility of such a domain. We present CompAS, a method that exploits the very evolution of innovations in the domain to carry out the necessary quantitative and qualitative commonality and variability analysis, especially in the case of scarce system documentation. We show how our technique applies to the CAOS domain by using conference proceedings as a key source of information about the evolution of features in CAOS systems over a period of several years. We detect and classify evolution patterns to determine functional commonality and variability. We also identify non-functional requirements to help capture domain variability. We have validated our approach by evaluating the degree to which representative test systems can be covered by the common and variable features produced by our analysis.

Testudinid Herpesviruses: A Review

Reptile medicine has been one of the fastest growing disciplines within the veterinary medicine arena during the last 20 yr. Infectious disease has proven to be one of the most interesting and challenging subspecialties of this discipline. Among the most significant pathogens discovered and investigated in the last 2 decades are the Testudinid herpesviruses, previously known as tortoise herpesviruses. The first article describing a bona fide Testudinid herpesvirus dates back to 30 yr ago. Several articles have followed and a number of features of these agents and of their associated diseases are now known. Nevertheless, several questions remain unanswered. The origin of the virus(es), the search for an effective therapy, the issue of the clinically healthy carrier and how to manage them, and the need to develop more-specific and sensitive diagnostic tests are just some of the “big” issues which will need to be tackled in the future. In this article we will review the major features of these viral agents, trying to provide a useful resource for veterinarians and researchers who either need to work with these viruses or simply to familiarize themselves with the topic.

Sparse representation based action and gesture recognition

In this paper we present a solution to the problem of action and gesture recognition using sparse representations. The dictionary is modelled as a simple concatenation of features computed for each action or gesture class from the training data, and test data is classified by finding sparse representation of the test video features over this dictionary. Our method does not impose any explicit training procedure on the dictionary. We experiment our model with two kinds of features, by projecting (i) Gait Energy Images (GEIs) and (ii) Motion-descriptors, to a lower dimension using Random projection. Experiments have shown 100% recognition rate on standard datasets and are compared to the results obtained with widely used SVM classifier.

Cochlear implant candidates with psychogenic hearing loss.

Abstract Conclusions: Specific requests for cochlear implantations by persons with psychogenic hearing loss are a relatively new phenomenon. A number of features seems to be over-represented in this group of patients. The existence of these requests stresses the importance of auditory brainstem response (ABR) measurements before cochlear implantation. Objective: To describe the phenomenon of patients with psychogenic hearing losses specifically requesting cochlear implantation, and to gain first insights into the characteristics of this group. Methods: Analysis of all cases seen between 2004 and 2013 at the University Hospital of Bern, Switzerland. Results: Four cochlear implant candidates with psychogenic hearing loss were identified. All were female, aged 23-51 years. Hearing thresholds ranged from 86 dB to 112 dB HL (pure-tone average 500-4000 Hz). ABRs and otoacoustic emissions (OAEs) showed bilaterally normal hearing in two subjects, and hearing thresholds between 30 and 50 dB in the other two subjects. Three subjects suffered from depression and one from a pathologic fear of cancer. Three had a history of five or more previous surgeries. Three were smokers and three reported other close family members with hearing losses. All four were hearing aid users at the time of presentation.