Content-based image retrieval using spatial layout information in brain tumor T1-weighted contrast-enhanced MR images

This study aims to develop content-based image retrieval (CBIR) system for the retrieval of T1-weighted contrast-enhanced MR (CE-MR) images of brain tumors. When a tumor region is fed to the CBIR system as a query, the system attempts to retrieve tumors of the same pathological category. The bag-of-visual-words (BoVW) model with partition learning is incorporated into the system to extract informative features for representing the image contents. Furthermore, a distance metric learning algorithm called the Rank Error-based Metric Learning (REML) is proposed to reduce the semantic gap between low-level visual features and high-level semantic concepts. The effectiveness of the proposed method is evaluated on a brain T1-weighted CE-MR dataset with three types of brain tumors (i.e., meningioma, glioma, and pituitary tumor). Using the BoVW model with partition learning, the mean average precision (mAP) of retrieval increases beyond 4.6% with the learned distance metrics compared with the spatial pyramid BoVW method. The distance metric learned by REML significantly outperforms three other existing distance metric learning methods in terms of mAP. The mAP of the CBIR system is as high as 91.8% using the proposed method, and the precision can reach 93.1% when the top 10 images are returned by the system. These preliminary results demonstrate that the proposed method is effective and feasible for the retrieval of brain tumors in T1-weighted CE-MR Images.

A coding method for efficient subgraph querying on vertex- and edge-labeled graphs

Labeled graphs are widely used to model complex data in many domains, so subgraph querying has been attracting more and more attention from researchers around the world. Unfortunately, subgraph querying is very time consuming since it involves subgraph isomorphism testing that is known to be an NP-complete problem. In this paper, we propose a novel coding method for subgraph querying that is based on Laplacian spectrum and the number of walks. Our method follows the filtering-and-verification framework and works well on graph databases with frequent updates. We also propose novel two-step filtering conditions that can filter out most false positives and prove that the two-step filtering conditions satisfy the no-false-negative requirement (no dismissal in answers). Extensive experiments on both real and synthetic graphs show that, compared with six existing counterpart methods, our method can effectively improve the efficiency of subgraph querying.

A novel harmony search algorithm based on teaching-learning strategies for 0-1 knapsack problems

To enhance the performance of harmony search (HS) algorithm on solving the discrete optimization problems, this paper proposes a novel harmony search algorithm based on teaching-learning (HSTL) strategies to solve 0-1 knapsack problems. In the HSTL algorithm, firstly, a method is presented to adjust dimension dynamically for selected harmony vector in optimization procedure. In addition, four strategies (harmony memory consideration, teaching-learning strategy, local pitch adjusting, and random mutation) are employed to improve the performance of HS algorithm. Another improvement in HSTL method is that the dynamic strategies are adopted to change the parameters, which maintains the proper balance effectively between global exploration power and local exploitation power. Finally, simulation experiments with 13 knapsack problems show that the HSTL algorithm can be an efficient alternative for solving 0-1 knapsack problems.

An open-source, mobile-friendly search engine for public medical knowledge

The World Wide Web has become an important source of information for medical practitioners. To complement the capabilities of currently available web search engines we developed FindMeEvidence, an open-source, mobile-friendly medical search engine. In a preliminary evaluation, the quality of results from FindMeEvidence proved to be competitive with those from TRIP Database, an established, closed-source search engine for evidence-based medicine.

Extracting tag hierarchies

Tagging items with descriptive annotations or keywords is a very natural way to compress and highlight information about the properties of the given entity. Over the years several methods have been proposed for extracting a hierarchy between the tags for systems with a "flat", egalitarian organization of the tags, which is very common when the tags correspond to free words given by numerous independent people. Here we present a complete framework for automated tag hierarchy extraction based on tag occurrence statistics. Along with proposing new algorithms, we are also introducing different quality measures enabling the detailed comparison of competing approaches from different aspects. Furthermore, we set up a synthetic, computer generated benchmark providing a versatile tool for testing, with a couple of tunable parameters capable of generating a wide range of test beds. Beside the computer generated input we also use real data in our studies, including a biological example with a pre-defined hierarchy between the tags. The encouraging similarity between the pre-defined and reconstructed hierarchy, as well as the seemingly meaningful hierarchies obtained for other real systems indicate that tag hierarchy extraction is a very promising direction for further research with a great potential for practical applications. Tags have become very prevalent nowadays in various online platforms ranging from blogs through scientific publications to protein databases. Furthermore, tagging systems dedicated for voluntary tagging of photos, films, books, etc. with free words are also becoming popular. The emerging large collections of tags associated with different objects are often referred to as folksonomies, highlighting their collaborative origin and the “flat” organization of the tags opposed to traditional hierarchical categorization. Adding a tag hierarchy corresponding to a given folksonomy can very effectively help narrowing or broadening the scope of search. Moreover, recommendation systems could also benefit from a tag hierarchy.

Spatial distribution of cosmetic-procedure businesses in two U.S. cities: a pilot mapping and validation study

Cosmetic procedures have proliferated rapidly over the past few decades, with over $11 billion spent on cosmetic surgeries and other minimally invasive procedures and another $2.9 billion spent on U.V. indoor tanning in 2012 in the United States alone. While research interest is increasing in tandem with the growth of the industry, methods have yet to be developed to identify and geographically locate the myriad types of businesses purveying cosmetic procedures. Geographic location of cosmetic-procedure businesses is a critical element in understanding the public health impact of this industry; however no studies we are aware of have developed valid and feasible methods for spatial analyses of these types of businesses. The aim of this pilot validation study was to establish the feasibility of identifying businesses offering surgical and minimally invasive cosmetic procedures and to characterize the spatial distribution of these businesses. We developed and tested three methods for creating a geocoded list of cosmetic-procedure businesses in Boston (MA) and Seattle (WA), USA, comparing each method on sensitivity and staff time required per confirmed cosmetic-procedure business. Methods varied substantially. Our findings represent an important step toward enabling rigorous health-linked spatial analyses of the health implications of this little-understood industry.

Management of scientific information with Google Drive

BACKGROUND

The amount and diversity of scientific publications requires a modern management system. By "management" we mean the process of gathering interesting information for the purpose of reading and archiving for quick access in future clinical practice and research activity. In the past, such system required physical existence of a library, either institutional or private. Nowadays in an era dominated by electronic information, it is natural to migrate entire systems to a digital form.

AIM OF THE STUDY

In the following paper we describe the structure and functions of an individual electronic library system (IELiS) for the management of scientific publications based on the Google Drive service.

MATERIAL AND METHODS

Architecture of the system. Architecture system consists of a central element and peripheral devices. Central element of the system is virtual Google Drive provided by Google Inc. Physical elements of the system include: tablet with Android operating system and a personal computer, both with internet access. Required software includes a program to view and edit files in PDF format for mobile devices and another to synchronize the files.

RESULTS

Functioning of the system. The first step in creating a system is collection of scientific papers in PDF format and their analysis. This step is performed most frequently on a tablet. At this stage, after being read, the papers are cataloged in a system of folders and subfolders, according to individual demands. During this stage, but not exclusively, the PDF files are annotated by the reader. This allows the user to quickly track down interesting information in review or research process. Modification of the document title is performed at this stage, as well. Second element of the system is creation of a mirror database in the Google Drive virtual memory. Modified and cataloged papers are synchronized with Google Drive. At this stage, a fully functional scientific information electronic library becomes available online. The third element of the system is a periodic two-way synchronization of data between Google Drive and tablet, as occasional modification of the files with annotation or recataloging may be performed at both locations.

CONCLUSIONS

The system architecture is designed to gather, catalog and analyze scientific publications. All steps are electronic, eliminating paper forms. Indexed files are available for re-reading and modification. The system allows for fast access to full-text search with additional features making research easier. Team collaboration is also possible with full control of user privileges. Particularly important is the safety of collected data. In our opinion, the system exceeds many commercially available applications in terms of functionality and versatility.

GRtoGR: a system for mapping GO relations to gene relations

We introduce in this paper a biological search engine called GRtoGR. Given a set of S genes, GRtoGR would determine from GO graph the most significant Lowest Common Ancestor (LCA) of the GO terms annotating the set S. This significant LCA annotates the genes that are the most semantically related to the set S. The framework of GRtoGR refines the concept of LCA by introducing the concepts of Relevant Lowest Common Ancestor (RLCA) and Semantically Relevant Lowest Common Ancestor (SRLCA). A SRLCA is the most significant LCA of the GO terms annotating the set S. We observe that the existence of the GO terms annotating the set S is dependent on the existence of this SRLCA in GO graph. That is, the terms annotating a given set of genes usually have existence dependency relationships with the SRLCA of these terms. We evaluated GRtoGR experimentally and compared it with nine other methods. Results showed marked improvement.

A scalable memetic algorithm for simultaneous instance and feature selection

Instance selection is becoming increasingly relevant due to the huge amount of data that is constantly produced in many fields of research. At the same time, most of the recent pattern recognition problems involve highly complex datasets with a large number of possible explanatory variables. For many reasons, this abundance of variables significantly harms classification or recognition tasks. There are efficiency issues, too, because the speed of many classification algorithms is largely improved when the complexity of the data is reduced. One of the approaches to address problems that have too many features or instances is feature or instance selection, respectively. Although most methods address instance and feature selection separately, both problems are interwoven, and benefits are expected from facing these two tasks jointly. This paper proposes a new memetic algorithm for dealing with many instances and many features simultaneously by performing joint instance and feature selection. The proposed method performs four different local search procedures with the aim of obtaining the most relevant subsets of instances and features to perform an accurate classification. A new fitness function is also proposed that enforces instance selection but avoids putting too much pressure on removing features. We prove experimentally that this fitness function improves the results in terms of testing error. Regarding the scalability of the method, an extension of the stratification approach is developed for simultaneous instance and feature selection. This extension allows the application of the proposed algorithm to large datasets. An extensive comparison using 55 medium to large datasets from the UCI Machine Learning Repository shows the usefulness of our method. Additionally, the method is applied to 30 large problems, with very good results. The accuracy of the method for class-imbalanced problems in a set of 40 datasets is shown. The usefulness of the method is also tested using decision trees and support vector machines as classification methods.

TrOn: an anatomical ontology for the beetle Tribolium castaneum

In a morphological ontology the expert's knowledge is represented in terms, which describe morphological structures and how these structures relate to each other. With the assistance of ontologies this expert knowledge is made processable by machines, through a formal and standardized representation of terms and their relations to each other. The red flour beetle Tribolium castaneum, a representative of the most species rich animal taxon on earth (the Coleoptera), is an emerging model organism for development, evolution, physiology, and pest control. In order to foster Tribolium research, we have initiated the Tribolium Ontology (TrOn), which describes the morphology of the red flour beetle. The content of this ontology comprises so far most external morphological structures as well as some internal ones. All modeled structures are consistently annotated for the developmental stages larva, pupa and adult. In TrOn all terms are grouped into three categories: Generic terms represent morphological structures, which are independent of a developmental stage. In contrast, downstream of such terms are concrete terms which stand for a dissectible structure of a beetle at a specific life stage. Finally, there are mixed terms describing structures that are only found at one developmental stage. These terms combine the characteristics of generic and concrete terms with features of both. These annotation principles take into account the changing morphology of the beetle during development and provide generic terms to be used in applications or for cross linking with other ontologies and data resources. We use the ontology for implementing an intuitive search function at the electronic iBeetle-Base, which stores morphological defects found in a genome wide RNA interference (RNAi) screen. The ontology is available for download at http://ibeetle-base.uni-goettingen.de.

An improved swarm optimization for parameter estimation and biological model selection

One of the key aspects of computational systems biology is the investigation on the dynamic biological processes within cells. Computational models are often required to elucidate the mechanisms and principles driving the processes because of the nonlinearity and complexity. The models usually incorporate a set of parameters that signify the physical properties of the actual biological systems. In most cases, these parameters are estimated by fitting the model outputs with the corresponding experimental data. However, this is a challenging task because the available experimental data are frequently noisy and incomplete. In this paper, a new hybrid optimization method is proposed to estimate these parameters from the noisy and incomplete experimental data. The proposed method, called Swarm-based Chemical Reaction Optimization, integrates the evolutionary searching strategy employed by the Chemical Reaction Optimization, into the neighbouring searching strategy of the Firefly Algorithm method. The effectiveness of the method was evaluated using a simulated nonlinear model and two biological models: synthetic transcriptional oscillators, and extracellular protease production models. The results showed that the accuracy and computational speed of the proposed method were better than the existing Differential Evolution, Firefly Algorithm and Chemical Reaction Optimization methods. The reliability of the estimated parameters was statistically validated, which suggests that the model outputs produced by these parameters were valid even when noisy and incomplete experimental data were used. Additionally, Akaike Information Criterion was employed to evaluate the model selection, which highlighted the capability of the proposed method in choosing a plausible model based on the experimental data. In conclusion, this paper presents the effectiveness of the proposed method for parameter estimation and model selection problems using noisy and incomplete experimental data. This study is hoped to provide a new insight in developing more accurate and reliable biological models based on limited and low quality experimental data.

Influenza forecasting with Google Flu Trends

BACKGROUND

We developed a practical influenza forecast model based on real-time, geographically focused, and easy to access data, designed to provide individual medical centers with advanced warning of the expected number of influenza cases, thus allowing for sufficient time to implement interventions. Secondly, we evaluated the effects of incorporating a real-time influenza surveillance system, Google Flu Trends, and meteorological and temporal information on forecast accuracy.

METHODS

Forecast models designed to predict one week in advance were developed from weekly counts of confirmed influenza cases over seven seasons (2004-2011) divided into seven training and out-of-sample verification sets. Forecasting procedures using classical Box-Jenkins, generalized linear models (GLM), and generalized linear autoregressive moving average (GARMA) methods were employed to develop the final model and assess the relative contribution of external variables such as, Google Flu Trends, meteorological data, and temporal information.

RESULTS

A GARMA(3,0) forecast model with Negative Binomial distribution integrating Google Flu Trends information provided the most accurate influenza case predictions. The model, on the average, predicts weekly influenza cases during 7 out-of-sample outbreaks within 7 cases for 83% of estimates. Google Flu Trend data was the only source of external information to provide statistically significant forecast improvements over the base model in four of the seven out-of-sample verification sets. Overall, the p-value of adding this external information to the model is 0.0005. The other exogenous variables did not yield a statistically significant improvement in any of the verification sets.

CONCLUSIONS

Integer-valued autoregression of influenza cases provides a strong base forecast model, which is enhanced by the addition of Google Flu Trends confirming the predictive capabilities of search query based syndromic surveillance. This accessible and flexible forecast model can be used by individual medical centers to provide advanced warning of future influenza cases.

Development of a semantic-based search system for immunization knowledge

This study developed and implemented a children's immunization management system with English and Traditional Chinese immunization ontology for semantic-based search of immunization knowledge. Parents and guardians are able to search vaccination-related information effectively. Jena Java Application Programming Interface (API) was used to search for synonyms and associated classes in this domain and then use them for searching by Google Search API. The searching results do not only contain suggested web links but also include a basic introduction to vaccine and related preventable diseases. Compared with the Google keyword-based search, over half of the 31 trial users prefer using semantic-based search of this system. Although the search runtime on this system is not as fast as well-known search engines such as Google or Yahoo, it can accurately focus on searching for child vaccination information to provide search results that better conform to the needs of users. Furthermore, the system is also one of the few health knowledge platforms that support Traditional Chinese semantic-based search.

Multi-lingual search engine to access PubMed monolingual subsets: a feasibility study

PubMed contains many articles in languages other than English but it is difficult to find them using the English version of the Medical Subject Headings (MeSH) Thesaurus. The aim of this work is to propose a tool allowing access to a PubMed subset in one language, and to evaluate its performance. Translations of MeSH were enriched and gathered in the information system. PubMed subsets in main European languages were also added in our database, using a dedicated parser. The CISMeF generic semantic search engine was evaluated on the response time for simple queries. MeSH descriptors are currently available in 11 languages in the information system. All the 654,000 PubMed citations in French were integrated into CISMeF database. None of the response times exceed the threshold defined for usability (2 seconds). It is now possible to freely access biomedical literature in French using a tool in French; health professionals and lay people with a low English language may find it useful. It will be expended to several European languages: German, Spanish, Norwegian and Portuguese.

Researchermap: a tool for visualizing author locations using Google maps

We hereby present ResearcherMap, a tool to visualize locations of authors of scholarly papers. In response to a query, the system returns a map of author locations. To develop the system we first populated a database of author locations, geocoding institution locations for all available institutional affiliation data in our database. The database includes all authors of Medline papers from 1990 to 2012. We conducted a formative heuristic usability evaluation of the system and measured the system's accuracy and performance. The accuracy of finding the accurate address is 97.5% in our system.

Search strategies to identify reports on "off-label" drug use in EMBASE

BACKGROUND

Medications are frequently prescribed outside their regulatory approval (off-label) by physicians particularly where appropriate therapies are not available. However, the risk/benefit ratio of drugs in off-label use needs to be critically appraised because it may differ from approved on-label usage. Therefore, an extensive exploration of current evidence on clinical data is well-advised. The objective of this study was to develop a search strategy that facilitates detection of the off-label drug use documents in EMBASE via OvidSP.

METHODS

We constructed two sets of gold standards from relevant records to off-label drug use by a sensitive search of MEDLINE and EMBASE. Search queries, including search words and strings, were conceived based on definition of off-label use of medications as well as text analysis of 500 randomly selected relevant documents. The selected terms were searched in EMBASE (from 1988 to 2011) and their retrieval performance was compared with the gold standards. We developed a sensitivity-maximizing, and a sensitivity- and precision-maximizing search strategy.

RESULTS

From 4067 records relevant to off-label drug use in our full gold standard set, 3846 records were retrievable from EMBASE. "off label*.af." was the most sensitive single term (overall sensitivity 77.5%, sensitivity within EMBASE 81.9%, precision 88.1%). The highest sensitive search strategy was achieved by combining 36 search queries with overall sensitivity of 94.0% and precision of 69.5%. An optimal sensitive and precise search strategy was yielded precision 87.4% at the expense of decreasing overall sensitivity to 89.4%.

CONCLUSION

We developed highly sensitive search strategies to enhance the retrieval of studies on off-label drug use in OvidSP EMBASE.

[Construction and application of natural products LC-MS-DS]

To overcome the defects of current research methods of natural products, an LC-MS-DS that consisted of more than 600 natural products was constructed with HPLC-ESI-IT-MS/MS. The database searching results of LC-MS-DS were validated to be reliable, which indicated that LC-MS-DS could be successfully applied for the effective identification and target isolation of natural products. The study will provide an effective method for fast identification and efficient isolation of natural products.

The performance of adverse effects search filters in MEDLINE and EMBASE

BACKGROUND

 Search filters can potentially improve the efficiency of searches involving electronic databases such as medline and embase. Although search filters have been developed for identifying records that contain adverse effects data, little is known about the sensitivity of such filters.

OBJECTIVES

 This study measured the sensitivity of using available adverse effects filters to retrieve papers with adverse effects data.

METHODS

  A total of 233 included studies from 26 systematic reviews of adverse effects were used for analysis. Search filters from medline and embase were tested for their sensitivity in retrieving the records included in these reviews. In addition, the sensitivity of each individual search term used in at least one search filter was measured.

RESULTS

 Subheadings proved the most useful search terms in both medline and embase. No indexing terms in medline achieved over 12% sensitivity. The sensitivity of published search filters varied in medline from 3% to 93% and in embase from 57% to 97%. Whether this level of sensitivity is acceptable will be dependent on the purpose of the search.

CONCLUSIONS

 Although no adverse effects search filter captured all the relevant records, high sensitivity could be achieved. Search filters may therefore be useful in retrieving adverse effects data.

Refining comparative proteomics by spectral counting to account for shared peptides and multiple search engines

Spectral counting has become a widely used approach for measuring and comparing protein abundance in label-free shotgun proteomics. However, when analyzing complex samples, the ambiguity of matching between peptides and proteins greatly affects the assessment of peptide and protein inventories, differentiation, and quantification. Meanwhile, the configuration of database searching algorithms that assign peptides to MS/MS spectra may produce different results in comparative proteomic analysis. Here, we present three strategies to improve comparative proteomics through spectral counting. We show that comparing spectral counts for peptide groups rather than for protein groups forestalls problems introduced by shared peptides. We demonstrate the advantage and flexibility of this new method in two datasets. We present four models to combine four popular search engines that lead to significant gains in spectral counting differentiation. Among these models, we demonstrate a powerful vote counting model that scales well for multiple search engines. We also show that semi-tryptic searching outperforms tryptic searching for comparative proteomics. Overall, these techniques considerably improve protein differentiation on the basis of spectral count tables.

BOSS: context-enhanced search for biomedical objects

BACKGROUND

There exist many academic search solutions and most of them can be put on either ends of spectrum: general-purpose search and domain-specific "deep" search systems. The general-purpose search systems, such as PubMed, offer flexible query interface, but churn out a list of matching documents that users have to go through the results in order to find the answers to their queries. On the other hand, the "deep" search systems, such as PPI Finder and iHOP, return the precompiled results in a structured way. Their results, however, are often found only within some predefined contexts. In order to alleviate these problems, we introduce a new search engine, BOSS, Biomedical Object Search System.

METHODS

Unlike the conventional search systems, BOSS indexes segments, rather than documents. A segment refers to a Maximal Coherent Semantic Unit (MCSU) such as phrase, clause or sentence that is semantically coherent in the given context (e.g., biomedical objects or their relations). For a user query, BOSS finds all matching segments, identifies the objects appearing in those segments, and aggregates the segments for each object. Finally, it returns the ranked list of the objects along with their matching segments.

RESULTS

The working prototype of BOSS is available at http://boss.korea.ac.kr. The current version of BOSS has indexed abstracts of more than 20 million articles published during last 16 years from 1996 to 2011 across all science disciplines.

CONCLUSION

BOSS fills the gap between either ends of the spectrum by allowing users to pose context-free queries and by returning a structured set of results. Furthermore, BOSS exhibits the characteristic of good scalability, just as with conventional document search engines, because it is designed to use a standard document-indexing model with minimal modifications. Considering the features, BOSS notches up the technological level of traditional solutions for search on biomedical information.