Detección de sonidos sibilantes en señales de audio respiratorias monocanal.
Archivos
NO SE HA AUTORIZADO la consulta de los documentos asociados
Fecha
2024-10-16
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
[ES] Los sonidos sibilantes son considerados uno de los biomarcadores acústicos más relevantes para la detección de patologías del sistema respiratorio humano, como el asma o la enfermedad pulmonar obstructiva crónica. En la última década, los enfoques basados en “Convolutional Neural Networks (CNN)” han demostrado ser los más relevantes para la detección de estos sonidos. Sin embargo, estos métodos requieren de una base de datos de sonidos reales extensa y variada para obtener un rendimiento óptimo. No obstante, existe una carencia significativa de bases de datos públicas de sonidos sibilantes que supone una limitación crucial en este ámbito de investigación. En este sentido, el objetivo principal de este Trabajo Fin de Grado es la generación de bases de datos sintéticas de sonidos sibilantes a partir de un modelado matemático del comportamiento fisiológico de estos sonidos y el desarrollo de herramientas basadas en CNN para la detección de sonidos sibilantes.
[EN] Wheezing sounds are considered one of the most relevant acoustic biomarkers for the detection of pathologies of the human respiratory system, such as asthma or chronic obstructive pulmonary disease. In the last decade, approaches based on convolutional neural networks (CNN) have demonstrated to be the most relevant for the detection of these sounds. However, these methods require a large and varied database of real sounds for optimal performance. Nevertheless, there is a significant lack of public databases of wheezing sounds, which is a crucial limitation in this area of research. In this sense, the main objective of this Final Degree Project is the generation of synthetic databases of wheezing sounds based on a mathematical modelling of the physiological behaviour of these sounds and the development of CNN-based tools for the detection of wheezing sounds.
[EN] Wheezing sounds are considered one of the most relevant acoustic biomarkers for the detection of pathologies of the human respiratory system, such as asthma or chronic obstructive pulmonary disease. In the last decade, approaches based on convolutional neural networks (CNN) have demonstrated to be the most relevant for the detection of these sounds. However, these methods require a large and varied database of real sounds for optimal performance. Nevertheless, there is a significant lack of public databases of wheezing sounds, which is a crucial limitation in this area of research. In this sense, the main objective of this Final Degree Project is the generation of synthetic databases of wheezing sounds based on a mathematical modelling of the physiological behaviour of these sounds and the development of CNN-based tools for the detection of wheezing sounds.
Descripción
Palabras clave
Especialidad en Sistemas de Telecomunicación y especialidad en Sonido e Imagen