Anthome Aalwan Vaz1, Mohammed Imran1, Muruganantham Sadasivam1, Syed Farhan Hassan1, Vigneshwaran Mahendiran1 and Parisa Naraei2*
1AIMT, Lambton College, Toronto, Canada
2Principal Investigator Applied Research Center, Lambton College, Canada
*Corresponding Author: Parisa Naraei, Principal Investigator Applied Research Center, Lambton College, Canada.
Received: April 26, 2022; Published: May 18, 2022
The initial phase of the project has been started with the data collection, where we have collected ECG images and 12 lead ECG waveform values of patients affected with covid 19 with five classifications as Covid 19, abnormal heart rate, myocardial infarction (MI), history of MI and normal ECG. We have done gamma correction for the image dataset in data preprocessing and cleaning, and waveforms have been created from 12 lead ECG values. To increase the available data, we have augmented the created images. We have created test and validation classes for the image dataset to train the model. A sequential convolutional neural network has been built for image classification. We used the batch normalization method from Keras; in the sequential model layer, we used relu and softmax activation layers—Adam optimizer and sparse categorical cross-entropy as the loss function in the sequential model. Training and validation accuracy have been used as metrics to assess the model's performance. Vgg16 model has been trained to compare the results. On completion of model creation, the trained model has been exported, and mobile application integration has been made for the end-user.
Keywords: Covid-19; ECG Image; Arrhythmia; Data Collection; Data Preprocessing; Data Cleaning; Data Augmentation; Model Building; Training; Model Evaluation; Android Application Development; CNN; Custom CNN; Vgg16
Citation: Parisa Naraei., et al. “Covid-19 and Cardiac Disease Classification using ECG Images". Acta Scientific Computer Sciences 4.6 (2022): 35-45.
Copyright: © 2022 Parisa Naraei., et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.