RESUMEN

Resumen: La evaluación automática de sonidos de auscultación cervical (AC) es una herramienta no invasiva para evaluación de la deglución. Sin embargo, los eventos deglutorios pueden verse enmascarados por fuentes de ruido. Este trabajo propone una metodología de caracterización y clasificación de señales de AC con alta resolución temporal a partir de estetoscopio, para discriminar entre sonidos deglutorios y asociados a ruido. Se adquirieron señales de AC en 10 sujetos sanos durante tres pruebas: toma de líquido, pronunciación del fonema /a/ y aclaramiento de garganta. Se extrajeron características de la señal de AC basadas en coeficientes cepstrales en la escala Mel, transformada wavelet discreta y entropía de Shannon. Las características con mayor relevancia fueron utilizadas como entrada a una máquina de vectores de soporte. Utilizando ventanas de 60 ms - alta resolución temporal - y validación cruzada, se obtuvieron exactitudes del 97.7% para detección de eventos acústicos y 91.7% para sonidos deglutorios. El método propuesto permite clasificación de sonidos deglutorios utilizando estetoscopio -dispositivo común en la práctica clínica- con exactitud comparable a otros trabajos que tienen menor resolución temporal o que utilizan otro tipo de sensores. Este trabajo constituye una primera etapa en el desarrollo de un algoritmo robusto para clasificación de sonidos deglutorios asociados a desórdenes de la deglución, a partir de auscultación cervical, para fines de diagnóstico automático.

Abstract: Automatic evaluation of cervical auscultation sounds (AC) is a non-invasive tool for swallowing assessment. However, the swallowing events could be perturbed by acoustic noise. This paper proposes a methodology of characterization and classification of AC signals acquired by stethoscope with high temporal resolution, in order to discriminate between swallowing sounds and other acoustic noise. AC signals from 10 healthy individuals were acquired with stethoscope during three tasks: liquid ingestion, phoneme /a/ pronunciation and throat clearing. Features based in Mel frequency cepstral coefficients, discrete wavelet transform and Shannon entropy, were extracted. Features with highest Fisher's discriminant ratio were used as input of a support vector machine. By application of 60 ms windows and cross validation, the obtained accuracies were 97.7% for acoustic event detection and 91.7% for swallowing sound detection. The proposed method allows classification swallowing sounds with higher temporal resolution­ than other works but with comparable accuracy. Furthermore, the use of stethoscope could lead to better acceptation than other sensors by physicians, because it is a common device in clinical practice. This work is a first stage in the development of a robust classification algorithm for sounds in swallowing disorders, oriented to automatic diagnosis.

RESUMEN

En los últimos años la comunidad científica internacional ha dedicado considerables recursos a la investigación y desarrollo de sistemas de diagnóstico asistidos por ordenador, utilizados por los médicos en el proceso de diagnóstico. Se ha prestado especial atención en algunas áreas médicas, como las especialidades oncológicas, por los altos índices de mortalidad provocados por algunas enfermedades como el cáncer de pulmón. El diagnóstico temprano de este padecimiento puede reducir en gran medida estos indicadores y mejorar la calidad de vida de los pacientes. El objetivo que se pretende con el desarrollo de esta investigación, es la selección adecuada de un algoritmo de clasificación, para ser utilizado en la fase que lleva el mismo nombre como parte de un sistema de diagnóstico asistido por ordenador para la clasificación de nódulos pulmonares solitarios. Para la selección adecuada del algoritmo de clasificación, se realiza un experimento utilizando las herramientas Weka v3.7.10 y Matlab 2013. Para determinar cuál de las técnicas estudiadas arroja mejores resultados de rendimiento, se utilizó el mismo conjunto de datos para las fases de entrenamiento, prueba y validación del clasificador, disponible en la base de datos internacional The Lung Image Database Consortium Image Collection(AU)

In recent years the international scientific community has devoted considerable resources to research and development of systems for computer-aided diagnosis used by physicians in the diagnostic process. Special attention has been provided in some medical areas, such as oncology specialties, by high mortality rates caused by some diseases like lung cancer. Early diagnosis of this condition can greatly reduce these indicators and improve quality of life of patients.The objective pursued with the development of this research is the proper selection of a classification algorithm, to be used in the phase that has the same name, as part of a system of computer-aided diagnosis for classification of solitary pulmonary nodules. For the selection of the appropriate classification algorithm, an experiment was performed using the tools Weka v3.7.10 and Matlab 2013. To determine which of the techniques studied produces better performance results, the same data set was used for the phases of training, testing and validation of the classifier, available in the international database The Lung Image Database Consortium Image Collection(AU)

Asunto(s)

RESUMEN

O câncer de mama é um tumor que se desenvolve como consequência de alterações genéticas em algum conjunto de células da mama, podendo atingir axilas e até mesmo outros órgãos. O diagnóstico é realizado por meio de informações coletadas através de exames e observações clínicas. Em sistemas informatizados tais informações são usadas como entrada para processos de auxílio ao diagnóstico. Objetivando aumentar a acurácia e precisão desses sistemas implementamos e comparamos dois classificadores de reconhecimento de padrões, nomeados: Naïve Bayes (NB) e Quadratic Discriminant Analysis (QDA). Para condução dos experimentos utilizamos uma base de dados de 569 instâncias com 30 atributos e um rótulo identificador entre benignas e malignas, extraídas de imagens digitalizadas. Os resultados experimentais demonstram que, embora ambos classificadores sejam promissores, o classificador QDA apresenta, na média, melhores taxas de acurácia e sensibilidade.

Breast cancer is a tumor that develops as a consequence of genetic alterations in a number of breast cancercells, can reach the armpits and even other organs. The diagnosis is made based on information collected through tests and clinical observations. In computerized systems such information is used as input to aid the diagnosis process. Aiming to increase the accuracy and precision of these systems implemented and compared two pattern recognition classifiers appointed: Naïve Bayes (NB) and Quadratic Discriminant Analysis (QDA). To conduct the experiments we used a database of 569 instances with 30 attributes and a label identifier between benign and malignant, extracted from scanned images. The experimental results show that although both classifiers are promising the classifier Quadratic Discriminant Analysis has, on average, better accuracy rates and sensitivity.

Asunto(s)

RESUMEN

Assessing pigs' welfare is one of the most challenging subjects in intensive pig farming. Animal vocalization analysis is a noninvasive procedure and may be used as a tool for assessing animal welfare status. The objective of this research was to identify stress conditions in piglets reared in farrowing pens through their vocalization. Vocal signals were collected from 40 animals under the following situations: normal (baseline), feeling cold, in pain, and feeling hunger. A unidirectional microphone positioned about 15 cm from the animals' mouth was used for recording the acoustic signals. The microphone was connected to a digital recorder, where the signals were digitized at the 44,100 Hz frequency. The collected sounds were edited and analyzed. The J48 decision tree algorithm available at the Weka(®) data mining software was used for stress classification. It was possible to categorize diverse conditions from the piglets' vocalization during the farrowing phase (pain, cold and hunger), with an accuracy rate of 81.12%. Results indicated that vocalization might be an effective welfare indicator, and it could be applied for assessing distress from pain, cold and hunger in farrowing piglets.