ECG Classification for Detecting ECG Arrhythmia Empowered with Deep Learning Approaches

Table 9

Comparison result of proposed CAA-TL model with literature.

Studies	Data augmentation	Dataset used	Method		Findings
Strodthoff et al. [2]	No	PTB -XL	ResNet and inception		-Predicted accuracy 89.8%; -Less accurate
Wasimuddin et al. [5]	No	ECG-ID	CAD and machine learning		-Predicted accuracy 98.5%; -Handcrafted
Vijayakumar et al. [6]	No	No	Feature extraction to remove noise		-Predicted accuracy 94.5%; -Handcrafted
Hsu et al. [14]	No	MIT-DB	AlexNet and ResNet		-Predicted accuracy 94.4%; -Fewer images
Acharya et al. [16]	No	PTB DB	CNN layers		-Accuracy 93.5% with noise and 95.22% without noise; Less accurate
Gaddam et al. [37]	No	MIT-DB	AlexNet		-Predicted accuracy 95.6%; -Only 1 approach was used with less accuracy
Reddy and khare [32]	No	UCI dataset	Rule-based fuzzy classifier and feature reduction		-Predicted accuracy 76.51%; -Handcrafted
Poudel et al. [34]	No	KVASIR dataset	CAD and machine learning		F1-score of 0.88; -Handcrafted
Siddique et al. [36]	No	Private	Fuzzy inference system, deep extreme machine learning, and ANN		87.05%, 92.45%, and 89.4%; -Handcrafted
Proposed CAA-TL model	Yes	MIT-BIH	Transfer LearningMethods	AlexNet	Accuracy (98.38%)
				SqueezeNet	Accuracy (90.08%)
				ResNet50	Accuracy (91%)