Algoritmo automático para la detección de síndrome de QT Largo (SQTL).
Date
2024
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Concepción
Abstract
El Síndrome de QT Largo (SQTL) o Long QT Syndrome en inglés, es una afección cardíaca, la cual se caracteriza por una prolongación del intervalo QT. La mayoría de las veces no es visible a simple vista en el ECG, por lo que para su detección se debe segmentar los intervalos QT del ECG. Los métodos automatizados ofrecen ventajas comparados con la detección manual en términos de repetibilidad absoluta de las mediciones, inmunidad de errores relacionados con la fatiga del observador, lapsus de atención y transcripción. Este trabajo busca obtener las mediciones automatizadas de los intervalos QT, para detección de SQTL en la base de datos PTB Diagnostic ECG Database de Physionet. La base de datos consiste en mediciones de registros ECG de pacientes con distintas anomalías de la señal. La motivación es contribuir con desarrollar algoritmos avanzados de diagnóstico para el SQTL, logrando una detección certera del punto Q y el final de la onda T en toda la señal, con el fin de poder obtener resultados válidos mediante un algoritmo automático. Se desarrolló un nuevo algoritmo en Matlab llamado “Kenay QT” en el que se calcula de forma automática la totalidad de los puntos Q y onda T en la señal completa y se obtienen los intervalos QT válidos a lo largo de toda la señal de cada uno de los registros procesados. “Kenay QT” se basa en la búsqueda de los puntos fiduciales en la señal ECG mediante “ Pan and Tompkins”. Se evaluó en los 549 registros de la base de datos el algoritmo “Kenay QT”, obteniendo una visualización gráfica de todos los puntos Q, onda T, y complejo QRS de los registros. Los resultados del intervalo QT son entregados en milisegundos. Al comparar el rendimiento obtenido con los datos publicados utilizando la misma base de datos, se obtuvo una puntuación final (la diferencia RMS entre los intervalos QT de referencia y los intervalos QT correspondientes) de 16,05 ms, mejor que la puntuación obtenida mediante un método automatizado de código abierto de Yuriy Chesnokov quien obtuvo una puntuación final de 17,33 ms. El algoritmo desarrollado demuestra tener mejor puntuación y rendimiento en la detección de intervalos QT, facilitando la detección del SQTL, cumpliendo con los objetivos planteados en esta memoria.
Long QT Syndrome (LQTS) is a cardiac condition characterized by a prolongation of the QT interval. Most of the time it is not visible to the naked eye on the ECG, so for its detection the QT intervals of the ECG must be segmented. Automated methods offer advantages compared to manual detection in terms of absolute repeatability of measurements, immunity from errors related to observer fatigue, lapses of attention and transcription. This work seeks to obtain automated QT interval measurements for LQTS detection from Physionet’s PTB Diagnostic ECG Database. The database consists of measurements of ECG recordings from patients with different signal abnormalities. The motivation is to contribute with developing advanced diagnostic algorithms for LQTS, achieving accurate detection of the Q-point and the end of the T-wave in the whole signal, in order to be able to obtain valid results by an automatic algorithm. A new algorithm was developed in Matlab called “Kenay QT” in which the totality of the Q points and T wave in the complete signal is automatically calculated and the valid QT intervals are obtained along the whole signal of each of the processed records. “Kenay QT” is based on the search of the fiducial points in the ECG signal by “ Pan and Tompkins”. The “Kenay QT” algorithm was evaluated on the 549 records in the database, obtaining a graphic display of all Q points, T wave, and QRS complex of the records. The QT interval results are delivered in milliseconds. Comparing the performance obtained with published data using the same database, a final score (the RMS difference between the reference QT intervals and the corresponding QT intervals) of 16.05 ms was obtained, better than the score obtained using an open source automated method by Yuriy Chesnokov who obtained a final score of 17.33 ms. The developed algorithm demonstrates better score and performance in the detection of QT intervals, facilitating LQTS detection, fulfilling the objectives set out in this report.
Long QT Syndrome (LQTS) is a cardiac condition characterized by a prolongation of the QT interval. Most of the time it is not visible to the naked eye on the ECG, so for its detection the QT intervals of the ECG must be segmented. Automated methods offer advantages compared to manual detection in terms of absolute repeatability of measurements, immunity from errors related to observer fatigue, lapses of attention and transcription. This work seeks to obtain automated QT interval measurements for LQTS detection from Physionet’s PTB Diagnostic ECG Database. The database consists of measurements of ECG recordings from patients with different signal abnormalities. The motivation is to contribute with developing advanced diagnostic algorithms for LQTS, achieving accurate detection of the Q-point and the end of the T-wave in the whole signal, in order to be able to obtain valid results by an automatic algorithm. A new algorithm was developed in Matlab called “Kenay QT” in which the totality of the Q points and T wave in the complete signal is automatically calculated and the valid QT intervals are obtained along the whole signal of each of the processed records. “Kenay QT” is based on the search of the fiducial points in the ECG signal by “ Pan and Tompkins”. The “Kenay QT” algorithm was evaluated on the 549 records in the database, obtaining a graphic display of all Q points, T wave, and QRS complex of the records. The QT interval results are delivered in milliseconds. Comparing the performance obtained with published data using the same database, a final score (the RMS difference between the reference QT intervals and the corresponding QT intervals) of 16.05 ms was obtained, better than the score obtained using an open source automated method by Yuriy Chesnokov who obtained a final score of 17.33 ms. The developed algorithm demonstrates better score and performance in the detection of QT intervals, facilitating LQTS detection, fulfilling the objectives set out in this report.
Description
Memoria de Titulo para optar al título profesional de Ingeniera/o Civil Biomédica/o
Keywords
Síndrome de QT prolongado, Electrocardiogramas, Algoritmos