Advancing Bag-of-visual-words Representations For Lesion Classification In Retinal Images.

Diabetic Retinopathy (DR) is a complication of diabetes that can lead to blindness if not readily discovered. Automated screening algorithms have the potential to improve identification of patients who need further medical attention. However, the identification of lesions must be accurate to be useful for clinical application. The bag-of-visual-words (BoVW) algorithm employs a maximum-margin classifier in a flexible framework that is able to detect the most common DR-related lesions such as microaneurysms, cotton-wool spots and hard exudates. BoVW allows to bypass the need for pre- and post-processing of the retinographic images, as well as the need of specific ad hoc techniques for identification of each type of lesion. An extensive evaluation of the BoVW model, using three large retinograph datasets (DR1, DR2 and Messidor) with different resolution and collected by different healthcare personnel, was performed. The results demonstrate that the BoVW classification approach can identify different lesions within an image without having to utilize different algorithms for each lesion reducing processing time and providing a more flexible diagnostic system. Our BoVW scheme is based on sparse low-level feature detection with a Speeded-Up Robust Features (SURF) local descriptor, and mid-level features based on semi-soft coding with max pooling. The best BoVW representation for retinal image classification was an area under the receiver operating characteristic curve (AUC-ROC) of 97.8% (exudates) and 93.5% (red lesions), applying a cross-dataset validation protocol. To assess the accuracy for detecting cases that require referral within one year, the sparse extraction technique associated with semi-soft coding and max pooling obtained an AUC of 94.2 ± 2.0%, outperforming current methods. Those results indicate that, for retinal image classification tasks in clinical practice, BoVW is equal and, in some instances, surpasses results obtained using dense detection (widely believed to be the best choice in many vision problems) for the low-level descriptors.

Sensitivity and specificity of machine learning classifiers for glaucoma diagnosis using Spectral Domain OCT and standard automated perimetry

PURPOSE: To evaluate the sensitivity and specificity of machine learning classifiers (MLCs) for glaucoma diagnosis using Spectral Domain OCT (SD-OCT) and standard automated perimetry (SAP). METHODS: Observational cross-sectional study. Sixty two glaucoma patients and 48 healthy individuals were included. All patients underwent a complete ophthalmologic examination, achromatic standard automated perimetry (SAP) and retinal nerve fiber layer (RNFL) imaging with SD-OCT (Cirrus HD-OCT; Carl Zeiss Meditec Inc., Dublin, California). Receiver operating characteristic (ROC) curves were obtained for all SD-OCT parameters and global indices of SAP. Subsequently, the following MLCs were tested using parameters from the SD-OCT and SAP: Bagging (BAG), Naive-Bayes (NB), Multilayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RAN), Ensemble Selection (ENS), Classification Tree (CTREE), Ada Boost M1(ADA),Support Vector Machine Linear (SVML) and Support Vector Machine Gaussian (SVMG). Areas under the receiver operating characteristic curves (aROC) obtained for isolated SAP and OCT parameters were compared with MLCs using OCT+SAP data. RESULTS: Combining OCT and SAP data, MLCs' aROCs varied from 0.777(CTREE) to 0.946 (RAN).The best OCT+SAP aROC obtained with RAN (0.946) was significantly larger the best single OCT parameter (p<0.05), but was not significantly different from the aROC obtained with the best single SAP parameter (p=0.19). CONCLUSION: Machine learning classifiers trained on OCT and SAP data can successfully discriminate between healthy and glaucomatous eyes. The combination of OCT and SAP measurements improved the diagnostic accuracy compared with OCT data alone.

Fault Detection in Power Distribution Systems Using Automated Integration of Computational Intelligence Tools

The main purpose of this paper is to present architecture of automated system that allows monitoring and tracking in real time (online) the possible occurrence of faults and electromagnetic transients observed in primary power distribution networks. Through the interconnection of this automated system to the utility operation center, it will be possible to provide an efficient tool that will assist in decisionmaking by the Operation Center. In short, the desired purpose aims to have all tools necessary to identify, almost instantaneously, the occurrence of faults and transient disturbances in the primary power distribution system, as well as to determine its respective origin and probable location. The compilations of results from the application of this automated system show that the developed techniques provide accurate results, identifying and locating several occurrences of faults observed in the distribution system.

Using different satellite imagery and classification techniques to assess the contribution of trees outside forests in the municipality of Maringá, Brazil

Forest cover of the Maringá municipality, located in northern Parana State, was mapped in this study. Mapping was carried out by using high-resolution HRC sensor imagery and medium resolution CCD sensor imagery from the CBERS satellite. Images were georeferenced and forest vegetation patches (TOFs - trees outside forests) were classified using two methods of digital classification: reflectance-based or the digital number of each pixel, and object-oriented. The areas of each polygon were calculated, which allowed each polygon to be segregated into size classes. Thematic maps were built from the resulting polygon size classes and summary statistics generated from each size class for each area. It was found that most forest fragments in Maringá were smaller than 500 m². There was also a difference of 58.44% in the amount of vegetation between the high-resolution imagery and medium resolution imagery due to the distinct spatial resolution of the sensors. It was concluded that high-resolution geotechnology is essential to provide reliable information on urban greens and forest cover under highly human-perturbed landscapes.

Comparison of methods for classification of the coefficient of variation in papaya

ABSTRACT The objective of this work was to study the distribution of values of the coefficient of variation (CV) in the experiments of papaya crop (Carica papaya L.) by proposing ranges to guide researchers in their evaluation for different characters in the field. The data used in this study were obtained by bibliographical review in Brazilian journals, dissertations and thesis. This study considered the following characters: diameter of the stalk, insertion height of the first fruit, plant height, number of fruits per plant, fruit biomass, fruit length, equatorial diameter of the fruit, pulp thickness, fruit firmness, soluble solids and internal cavity diameter, from which, value ranges were obtained for the CV values for each character, based on the methodology proposed by Garcia, Costa and by the standard classification of Pimentel-Gomes. The results obtained in this study indicated that ranges of CV values were different among various characters, presenting a large variation, which justifies the necessity of using specific evaluation range for each character. In addition, the use of classification ranges obtained from methodology of Costa is recommended.

Access to information: assessment of the use of automated interaction technologies in call centers

With the purpose of at lowering costs and reendering the demanded information available to users with no access to the internet, service companies have adopted automated interaction technologies in their call centers, which may or may not meet the expectations of users. Based on different areas of knowledge (man-machine interaction, consumer behavior and use of IT) 13 propositions are raised and a research is carried out in three parts: focus group, field study with users and interviews with experts. Eleven automated service characteristics which support the explanation for user satisfaction are listed, a preferences model is proposed and evidence in favor or against each of the 13 propositions is brought in. With balance scorecard concepts, a managerial assessment model is proposed for the use of automated call center technology. In future works, the propositions may become verifiable hypotheses through conclusive empirical research.

Diabetes como causa básica ou associada de morte no Estado de São Paulo, Brazil, 1992

INTRODUÇÃO: As estatísticas de mortalidade, em geral, baseiam-se na análise das causas básicas de óbito. No caso do diabetes, sua importância é sempre subestimada, pois os diabéticos geralmente morrem devido às complicações crônicas da doença, sendo estas que figuram como a causa básica do óbito. Para atenuar esse problema, deveriam ser analisadas todas as causas mencionadas no atestado de óbito. Como contribuição ao problema foi analisada a freqüência das menções do diabetes nas declarações de óbito e as principais causas associadas. METODOLOGIA: Os coeficientes específicos e a mortalidade proporcional por diabetes, como causa básica ou associada, foram calculados com base nas informações extraídas dos atestados de óbito, através do sistema ACME (Automated Classification of Medical Entities), para o Estado de São Paulo, em1992. RESULTADOS E CONCLUSÕES: De um total de 202.141 óbitos, o diabetes foi mencionado em 13.786 (6,8%), sendo a causa básica em 5.305 (2,6%). A proporção foi maior para mulheres do que para homens (10,1 vs 4,6% como causa mencionada e 6,1 vs 2,9% como causa básica). Entre os óbitos com menção de diabetes no atestado, as principais causas básicas foram: diabetes (38,5%), doenças cardiovasculares (37,2%), doenças respiratórias (8,5%) e neoplasias (4,8%). Quando o diabetes foi a causa básica, as principais causas associadas foram: doenças cardiovasculares (42,2%), respiratórias (10,7%) e geniturinárias (10,1%) . Nos casos onde o diabetes figura como causa associada, as principais causas básicas foram as doenças cardiovasculares (60,5%), respiratórias (13,8%) e neoplásicas (7,9%). Apesar das limitações dos dados obtidos dos atestados de óbito, observou-se que o diabetes representa uma importante causa de morte, traduzindo um problema de saúde de grande magnitude. Também, a análise pelas causas múltiplas de morte fornece um perfil da morbidade associada ao diabetes por ocasião do óbito, salientando a importância das doenças cardiovasculares.

Equivalência entre revisões da Classificação Internacional de Doenças: causas de morte

OBJETIVO: Avaliar a comparabilidade entre a causa básica e as causas múltiplas de morte codificadas segundo as regras e disposições correlatas da nona e da décima revisões da Classificação Internacional de Doenças. MÉTODOS: Os dados provieram de uma amostra sistemática de 3.313 declarações de óbito de falecidos residentes no Estado de São Paulo, no ano de 1992 (1,6% do total dos óbitos naquele ano). Os dados foram processados pelo sistema "Automated Classification of Medical Entities", incluindo códigos para todas as afecções mencionadas nos atestados médicos e a causa básica que havia sido avaliada e revista segundo as disposições da nona revisão. Todas as afecções foram recodificadas segundo as disposições da décima revisão e os códigos resultantes introduzidos no banco de dados original para seleção da causa básica pelo sistema de declarações de óbito de São Paulo. As tabulações das causas múltiplas de morte codificadas pela nona e pela décima revisões foram obtidas pelas versões respectivas do programa "Tabulador de Causas Múltiplas". A comparação das causas de morte foi realizada a partir dos capítulos de ambas as revisões da Classificação Internacional de Doenças. RESULTADOS/CONCLUSÕES: As mudanças mais importantes para as causas básicas, ocorridas nos capítulos I, III e VIII da nona revisão e nos correspondentes capítulos I, IV e X da décima revisão, devem-se ao deslocamento das mortes causadas pela doença devido ao vírus da imunodeficiência humana e pela preterição das pneumonias como causa de morte. Em relação às causas múltiplas de morte, verificou-se o aumento de menções de doenças respiratórias e a correspondente diminuição de menções incluídas no capítulo das afecções mal definidas, devido à recodificação da insuficiência respiratória.

Mortalidade feminina em idade reprodutiva no Estado de São Paulo, Brasil, 1991-1995: causas básicas de óbito e mortalidade materna

OBJETIVO: Descrever a mortalidade materna no período reprodutivo (15 a 49 anos) no Estado de São Paulo, de 1991 a 1995, segundo grupos etários e causas básicas de óbito. MÉTODOS: Foi fornecida pela Fundação Seade a listagem dos óbitos, com as causas básicas codificadas pela Classificação Internacional de Doenças, 9ª Revisão, utilizando-se o programa "Automated Classification of Medical Entities", as estimativas da população feminina segundo grupos etários e os números de nascidos vivos. Foram calculados coeficientes específicos por 100.000 mulheres, mortalidade materna por 100.000 nascidos vivos e percentagens de óbitos por subgrupos. Foram calculadas medianas dos coeficientes do quinquênio, para comparação das principais causas agrupadas nos capítulos. RESULTADOS: De 1991 a 1995 houve aumento da mortalidade por deficiência da imunidade celular a partir de 25 anos, parecendo traduzir um paralelismo com a curva ascendente da epidemia de AIDS em mulheres. Lesões e envenenamentos predominam nas mais jovens, porém a partir de 35 anos as doenças do aparelho circulatório e neoplasmas passaram a ser preponderantes. Doenças infecciosas e parasitárias ocupam a sétima ou oitava posição, em todas as idades. Acidentes e homicídios e suicídios foram elevados. A mortalidade materna variou de 43,7 a 49,6 por 100.000 nascidos vivos. CONCLUSÕES: Houve grande exposição das mulheres em idade fértil a fatores associados a causas externas, doenças crônicas e AIDS. A maioria das causas apontadas de mortalidade materna podem ser prevenidas e, portanto, revelam insuficiência de assistência pré-natal adequada e extensiva, bem como deficiências no atendimento ao parto e puerpério.

Comparison between the HemoCue® and an automated counter for measuring hemoglobin

To determine the precision and agreement of the hemoglobin (Hb) measurements in capillary and venous blood samples by the HemoCue® and an automated counter. Hb was determined by both equipaments in blood samples of 29 pregnant women. The HemoCue® showed low repeatability of Hb measurements in duplicate in capillary (CR=0.53 g/dL, CV=13.6%) and venous blood (CR=0.53 g/dL, CV=13.6%). Hb measurements in capillary blood were higher than those in venous blood (12.4 and 11.7 g/dL, respectively; p<0.05). There was high agreement between Hb in capillary blood by the HemoCue® and in venous blood by the counter (r icc=0.86; p<0.01), and also between the diagnosis of anemia by both equipments (k=0.81; p<0.01). The HemoCue® seems to be more appropriate for capillary blood and require training of the measurers.

Data mining framework for fatty liver disease classification in ultrasound: a hybrid feature extraction paradigm

PURPOSE: Fatty liver disease (FLD) is an increasing prevalent disease that can be reversed if detected early. Ultrasound is the safest and ubiquitous method for identifying FLD. Since expert sonographers are required to accurately interpret the liver ultrasound images, lack of the same will result in interobserver variability. For more objective interpretation, high accuracy, and quick second opinions, computer aided diagnostic (CAD) techniques may be exploited. The purpose of this work is to develop one such CAD technique for accurate classification of normal livers and abnormal livers affected by FLD. METHODS: In this paper, the authors present a CAD technique (called Symtosis) that uses a novel combination of significant features based on the texture, wavelet transform, and higher order spectra of the liver ultrasound images in various supervised learning-based classifiers in order to determine parameters that classify normal and FLD-affected abnormal livers. RESULTS: On evaluating the proposed technique on a database of 58 abnormal and 42 normal liver ultrasound images, the authors were able to achieve a high classification accuracy of 93.3% using the decision tree classifier. CONCLUSIONS: This high accuracy added to the completely automated classification procedure makes the authors' proposed technique highly suitable for clinical deployment and usage.

Comorbidity adjustment index for the international classification of diseases, 10

OBJECTIVE: To develop a Charlson-like comorbidity index based on clinical conditions and weights of the original Charlson comorbidity index. METHODS: Clinical conditions and weights were adapted from the International Classification of Diseases, 10th revision and applied to a single hospital admission diagnosis. The study included 3,733 patients over 18 years of age who were admitted to a public general hospital in the city of Rio de Janeiro, southeast Brazil, between Jan 2001 and Jan 2003. The index distribution was analyzed by gender, type of admission, blood transfusion, intensive care unit admission, age and length of hospital stay. Two logistic regression models were developed to predict in-hospital mortality including: a) the aforementioned variables and the risk-adjustment index (full model); and b) the risk-adjustment index and patient's age (reduced model). RESULTS: Of all patients analyzed, 22.3% had risk scores >1, and their mortality rate was 4.5% (66.0% of them had scores >1). Except for gender and type of admission, all variables were retained in the logistic regression. The models including the developed risk index had an area under the receiver operating characteristic curve of 0.86 (full model), and 0.76 (reduced model). Each unit increase in the risk score was associated with nearly 50% increase in the odds of in-hospital death. CONCLUSIONS: The risk index developed was able to effectively discriminate the odds of in-hospital death which can be useful when limited information is available from hospital databases.

Wetland Habitat Studies using various Classification Techniques on Multi-Spectral Landsat Imagery: Case study: Tram chim National Park, Dong Thap Vietnam

Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies

Automated illustration of multimedia stories

Submitted in part fulfillment of the requirements for the degree of Master in Computer Science

Automated image tagging through tag propagation

Trabalho apresentado no âmbito do Mestrado em Engenharia Informática, como requisito parcial Para obtenção do grau de Mestre em Engenharia Informática