Main Article Content
Time matters the most in medical diagnosis. Therefore, ECG has been extensively used in diagnosis and precise identification of several cardiac conditions. As continuation to the previous work, which concerned about feature extraction and classification of ECG signals applying CNN on distinct datasets, this paper presents residual learning model that makes training of networks easier and substantially deeper since the deeper neural networks are difficult to train. The proposed solution is classified into two types, one with no residual
connection (Basic CNN) and the other analyzing the repetition of residual connections (Residual CNN) for displaying the variation in performance level of each model. The proposed model offers extensive observational proof indicating that residual model is an optimized framework that shows better results of accuracy from considerably increased depth. From the performance of the (Deep Residual CNN) we also see that we can better exploit these residual connections by making the model much deeper. Such a model can be conveniently used for constant monitoring of ECG in real time environment.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.