DETECCIÓN DE ARRITMIAS EN ECG EMPLEANDO MÉTODOS DE APRENDIZAJE PROFUNDO (ECG ARRHYTHMIA DETECTION EMPLOYING DEEP LEARNING METHODS)
Resumen
Las enfermedades cardiovasculares son la principal causa de muerte en todo el mundo [OMS, 2021]. En este trabajo se presenta un sistema de Detección Asistida por Computadora (CADe) diseñado para identificar y clasificar anomalías en señales electrocardiográficas (ECG). Este sistema consta de dos etapas principales. (i) Se procesa la señal ECG, la cual se convierte en imágenes 2D utilizando la técnica de Recurrence Plot (RP), a partir de dividir la señal en segmentos de dos segundos. (ii) Las imágenes generadas se clasificaron utilizando la arquitectura ResNET-18 y el Módulo de Atención Convolucional (CBAM). La base de datos utilizada MIT-BIH, está compuesta por 16 clases de anomalías, empleando el estándar de la Asociación para el Avance de Instrumentación Médica (AAMI) se procede a agrupar las 16 anomalías en cinco categorías de relevancia médica. El sistema diseñado obtiene un rendimiento destacado, con una exactitud del 94.2%, precisión del 93.25%, recall del 93.24%, F1-Score del 93.24%.
Palabras Clave: CBAM; Electrocardiograma (ECG); Recurrence plot (RP); ResNet-18; y Random Under Sampling (RUS).
Abstract:
Cardiovascular diseases are the leading cause of death worldwide [OMS, 2021]. In the following work, a Computer-Aided Detection (CADe) system is presented, designed to identify, and classify anomalies in electrocardiographic (ECG) signals. This system consists of two main stages. (i) The ECG signal is processed, which is converted into 2D images using the Recurrence Plot (RP) technique, by dividing the signal into two-second segments. (ii) The generated images were classified using the ResNET-18 architecture and the Convolutional Block Attention Module (CBAM). The MIT-BIH database, composed of 16 classes of anomalies, was used, and following the standard of the Association for the Advancement of Medical Instrumentation (AAMI), the 16 anomalies were grouped into just five categories of medical relevance. The designed system achieves outstanding performance, with an accuracy of 94.2%, precision of 93.25%, recall of 93.24%, and F1-Score of 93.24%.
Keywords: CBAM; Electrocardiogram (ECG); Recurrence Plot (RP); ResNet-18, Random Under Sampling (RUS).
Texto completo:
596-614 PDFReferencias
AAMI, (1998). Standard Cardiac Arrhythmia Detection. IEEE, 16th Indian Council International Conference, No. 2, Vol. 5, pp.1-4.
Ahmad, Z., Tabassum, A., Guan, L., & Khan, N. M., (2021). ECG Heartbeat c.
Classification Using Multimodal Fusion, (2022). IEEE Access, No. 24, Vol. 9, pp. 100615-100626.
Eckmann, J. P., Kamphorst, S. O., & Ruelle, D., (1995). Recurrence Plots of Dynamical Systems. World Scientific Series on Nonlinear Science, Series A, No. 16, Vol. 16, pp. 441-446.
Faria, F. A., Almeida, J., Alberton, B., Morellato, L. P. C., & Torres, R. D. S. (2016), Fusion of Time Series Representations for Plant Recognition in Phenology Studies. Pattern Recognition Letters, No. 55, Vol. 83, pp. 205-214.
García, V., Mollineda, R. A., & Sánchez, J. S., (2009). Index of Balanced Accuracy: A Performance Measure for Skewed Class Distributions. In Iberian conference on pattern recognition and image analysis, No. 2321, Vol. 5524, pp. 441-448.
Han, J., Sun, G., Song, X., Zhao, J., Zhang, J., & Mao, Y., (2022). Detecting ECG Abnormalities Using an Ensemble Framework Enhanced by Bayesian Belief Network. Biomedical Signal Processing and Control, No. 225, Vol. 72, pp. 103320.
He, K., Zhang, X., Ren, S., & Sun, J., (2016). Identity Mappings in Deep Re-sidual Networks. In Computer Vision–ECCV 2016: 14th European Confer-ence, Amsterdam, The Netherlands, October 11–14, Proceedings, Part IV, No. 54, Vol. 14, pp. 630-645.
Jana, G. C., Agrawal, A., Pattnaik, P. K., & Sain, M., (2022). DWT-EMD Fea-ture Level Fusion-Based Approach Over Multi and Single Channel EEG Signals for Seizure Detection. Diagnostics, No. 21, Vol. 12, pp. 324-334.
Johnson, J. M., & Khoshgoftaar, T. M., (2019). Survey on Deep Learning with Class Imbalance. Journal of Big Data, No. 1, Vol. 6, pp. 1-54.
Mathunjwa, B. M., Lin, Y. T., Lin, C. H., Abbod, M. F., Sadrawi, M., & Shieh, J. S., (2022). ECG Recurrence Plot-Based Arrhythmia Classification Using Two-Dimensional Deep Residual CNN features. Sensors, No. 4, Vol. 22, pp. 1660.
Niken, Martono P., Nishiguchi, T., & Ohwada, H., (2022). ECG Signal Clas-sification Using Recurrence Plot-Based Approach and Deep Learning for Arrhythmia Prediction. In Asian Conference on Intelligent Information and Database Systems, No. 12, Vol. 3, pp. 327-335.
Organización Mundial de la Salud (OMS), (2021). The Burden of Cardio-vascular Disease. No. 23, Vol. 46, pp. 54-86.
Pan, J., & Tompkins, W. J., (1985). A Real-Time QRS Detection Algo-rithm. IEEE transactions on biomedical engineering. No. 42, Vol. 3, pp. 230-236.
PhysioNet, (1980). The impact of the MIT-BIH Arrytmia Database. IEEE Eng in Med and Biol, No. 3, Vol. 20, pp: 45-50.
Secretaria de salud, (2022). Cada año, 220 mil Personas Fallecen Debido a Enfermedades del Corazón. Ciudad de México, México.
Vélez Rodríguez, (2013). Pautas de la Electrocardiografía (2ª ed.), MAR-BÁN Libros, Cataluña, España.
Woo, S., Park, J., Lee, J. Y., & Kweon, I. S., (2018). Cbam: Convolutional block attention module. Proceedings of the European conference on com-puter vision (ECCV), No. 17, Vol. 39, pp. 3-19.
Yang, S., Orellana-Barrios, M., & Nugent, K., (2018). Research Methods: Clinical studies based on routine laboratory tests. The Southwest Respira-tory and Critical Care Chronicles, No. 23, Vol. 6, pp. 33-39.
URL de la licencia: https://creativecommons.org/licenses/by/3.0/deed.es
Pistas Educativas está bajo la Licencia Creative Commons Atribución 3.0 No portada. 
TECNOLÓGICO NACIONAL DE MÉXICO / INSTITUTO TECNOLÓGICO DE CELAYA
Antonio García Cubas Pte #600 esq. Av. Tecnológico, Celaya, Gto. México
Tel. 461 61 17575 Ext 5450 y 5146
pistaseducativas@itcelaya.edu.mx