Table 1.
Literature reviews using ECG for detection of normal, MI, CAD, and CHF.
| Author (Year) | Database | Feature Extraction Method (Classifiers) | Intra-Patient | Inter-Patient |
|---|---|---|---|---|
| Normal and CAD | ||||
| Acharya et al., (2017) [8] | Fantasia and St.Petersburg databases | HOS (KNN/DT) | ACC = 98.99% | |
| SEN = 97.75% | ||||
| SPE = 99.39%. | ||||
| Kumar et al., (2017) [9] | Fantasia and St.Petersburg databases | FAWT (LS-SVM) | ACC = 99.60% | |
| SEN = 99.57% | ||||
| SPE = 99.61% | ||||
| Normal and MI | ||||
| Baloglu et al., (2019) [10] | PTB diagnostic ECG database | CNN (Softmax) | ACC = 99.78% | |
| Han et al., (2019) [11] | PTB diagnostic ECG database | Energy entropy based on MODWPT; Feature fusion (SVM) | ACC = 99.75% | ACC = 92.69% |
| SEN = 99.37% | SEN = 80.96% | |||
| PPV = 99.70% | PPV = 86.14% | |||
| Sharma et al., (2018) [12] | PTB diagnostic ECG database | Wavelet decomposition based on biorthogonal filter bank, fuzzy entropy (KNN) | ACC = 99.62% | |
| SEN = 99.76% | ||||
| SPE = 99.12% | ||||
| Acharya et al., (2017) [13] | PTB diagnostic ECG database | 11-layer CNN (Softmax) | ACC = 95.22% | |
| SEN = 95.49% | ||||
| SPE = 94.19% | ||||
| Reasat et al., (2017) [14] | PTB diagnostic ECG database | CNN with inception block (Softmax) | ACC = 84.54% | |
| SEN = 85.33% | ||||
| SPE = 84.09% | ||||
| Sharma et al., (2017) [15] | PTB diagnostic ECG database | SWT Sample entropy, log energy entropy, and median slope; (SVM/KNN) | ACC = 98.84% | ACC = 81.71% |
| SEN = 99.35% | SEN = 79.01% | |||
| SPE = 98.29% | SPE = 79.26% | |||
| Padhy et al., (2017) [16] | PTB diagnostic ECG database | SVD (SVM) | ACC = 95.30% | |
| SEN = 94.60% | ||||
| SPE = 96.00% | ||||
| Acharya et al., (2016) [17] | PTB diagnostic ECG database | DWT (KNN) | ACC = 98.8% | |
| SEN = 99.45% | ||||
| SPE = 96.27% | ||||
| Normal and CHF | ||||
| Acharya et al., (2019) [18] | MITBIH Normal Sinus Rhythm, BIDMC CHF database | 1D-CNN (Softmax) | ACC = 98.97% | |
| SEN = 98.87% | ||||
| SPE = 99.01% | ||||
| Sudarshan et al., (2017) [19] | MIT-BIH Normal Sinus Rhythm Database, BIDMC CHF database | Dual tree complex wavelet transform (KNN) | ACC = 99.86% | |
| SEN = 99.78% | ||||
| SPE = 99.94% | ||||
| Subasi et al., (2013) [20] | BIDMC CHF database, MIT-BIH Arrhythmia database | Autoregressive (AR) Burg (C4.5 DT) | SEN = 99.77% | |
| SPE = 99.93% | ||||
| Normal, CAD and MI | ||||
| Acharya et al., (2017) [21] | St.Petersburgdatabases, PTB diagnostic ECG database, | DWT EMD DCT (KNN) |
ACC = 98.5% | |
| SEN = 98.5% | ||||
| SPE = 99.7% | ||||
| Normal, CAD, MI, and CHF | ||||
| Fujita et al., (2017) [22] | St.Petersburg databases, PTB diagnostic ECG database, BIDMC CHF database | WPD ReliefF (KNN) |
ACC =97.98% | |
| SEN = 99.61% | ||||
| SPE = 94.84% | ||||
| Acharya et al., (2017) [23] | St.Petersburg databases, PTB diagnostic ECG database, BIDMC CHF database | CWT Contourlet Transform Shearlet Transform (DT KNN) |
ACC = 99.55% | |
| SEN = 99.93% | ||||
| SPE = 99.24% | ||||
ACC: Accuracy, SEN: Sensitivity, SPE: Specificity, HOS: Higher-Order Statistics and Spectra, PCA: Principle Component Analysis, SVD: Singular Value Decomposition, LS-SVM: Least Squares Support Vector Machine, DWT: Discrete Wavelet Transform, FAWT: Flexible Analytic Wavelet Transform, SWT: Stationary Wavelet Transform, DCT: Discrete Cosine Transform, CWT: Continuous Wavelet Transform, EMD: Empirical Mode Decomposition, DT: Decision Tree, KNN: K-Nearest Neighbors, CNN: Convolution Neural Network.