Rapid detection of cardiac pathologies by neural networks using ECG signals (1D) and sECG images (3D): exploratory study with 6-channel ECG

Normally, the detection of cardiac pathologies is performed using one-dimensional electrocardiogram signals or 2D (two-dimensional) images. When working with electrocardiogram signals, they can be represented in the time and frequency domain (1D signals). However, this technique can present difficulties, such as the high cost of private health services or the time taken by the public health system to refer the patient to a cardiologist. In addition, the variety of cardiac pathologies (more than 20 types) is a problem in the diagnosis of the disease. On the other hand, one of the little-explored techniques for this diagnosis is surface electrocardiography (sECG). sECGs are 3D images (two dimensions in space and one in time). First, a 6-channel electrocardiograph was built to record the precordial signals of the heart. Subsequently, two models, LSTM and ResNet34 NN, were developed and showed high accuracy, 98.71%, and 93.65%, respectively. Measurements were performed on two volunteer patients for which both models were successful. The present study proposes the basis for developing a Decision Support Software (DSS) based on machine learning models.

• Главный автор: Aguiar Salazar, Evelyn Dayana (author)
• Формат: bachelorThesis
• Язык: eng
• Опубликовано: 2021
• Предметы: Imágenes sECG; Redes neuronales; Electrocardiógrafo 6-ch; MATLAB; Python; sECG images; Neural networks; 6-channels electrocardiograph
• Online-ссылка: http://repositorio.yachaytech.edu.ec/handle/123456789/451